Unless, of course, I don’t get it.

(The first result got fixed, this version (courtesy Google’s cache) has the goodness.)

I was resigned by this time to losing all my ratings and play counts, as it seemed that the only method that would work was to delete the existing library and recreate it from a hack of the XML version of the library file.  This was what I started when I got back from holiday.

I backed everything up, then edited the XML file and search-and-replaced the file paths in the entries with the new location.  Then, I deleted the “iTunes Library” file and started iTunes.  Sure enough, empty library.  I selected “Import…” and pointed iTunes at the XML file.  After a fair amount of grinding, it gave me an error saying that “not all the songs could be added as they could not be found”.  Sure enough, only about 1400 of nearly 2600 made it into the library.  My search-and-replace must have been incomplete…

I figured that rather than scratch the library yet again and try and find the 1200-odd entries in the XML file that I had failed to edit correctly, I’d just use “Add to Library…” to bring in the missing files.  This worked correctly, and my iTunes listing was back to its former glory (without play counts and ratings though).

Then the moment of truth — iPod time.  I had not plugged my iPod in at all after the library had blown up, so I was a bit nervous about the possibility of an iPod sync removing all the media and details from the iPod.  I was hoping though that iTunes would read the ratings and play counts off the iPod and restore them to the library.

I sat the iPod in its dock, and got an error about the iPod being synced to a different library.  Nuts.  I reopened the XML file and found a field called “Library Persistent ID” which was different between the original XML file and the new one — so unfortunately iTunes didn’t import the ID from the original file.

At this point I got a little blase about the whole thing.  I’d already trashed the library once, how much worse could it get?  I decided that this ID value had to be in the binary library file somewhere, so went looking for a binary editor that could help.  I found a very cool little program (with an equally cool name) called 0xED that allowed me to edit the iTunes library file — and lo and behold, right there in about the first 100 bytes of the file, was exactly the hex string that was in the XML file.  I shut down iTunes and substituted the old value into the binary library file, and restarted iTunes.  So far so good.  I plugged in the iPod…  and it synced!  The majority of the files were fine, there were a couple of podcast episodes and a few old miscellaneous files that had to be copied to the iPod.  Unfortunately though, the ratings and play counts didn’t come out of the iPod so are presumably gone for good.

So I’m working again, right?  Wrong.  This mess was the kick in the pants I needed to go away from using iTunes for music and switch to something Linux.  The main issue I have is with podcasts — I’ve never really been happy with the Juice/iTunes combination, since Juice splattered podcasts all over my library (it doesn’t put everything neatly into the “Podcasts” directory like iTunes did) — and I forgot the reason I switched from iTunes to Juice for podcatching in the first place.  I’ve got an idea in mind to combine a command-line cron-scheduled podcatcher with the MySQL backend for amaroK, but more on that when (if) I get something working.

I’m sure that in the past when I’ve relocated the library all I needed to do was select the new location in iTunes Preferences.  iTunes would recognise all the music in the new location and update the library.  This time, however, nothing I did would tell iTunes to look in the new location!  If I made the old location unavailable, it complained that all the music files were not locatable.  If the old location was present, iTunes just didn’t bother making any change to the index.  At one stage I told it to “Consolidate Library…”, one of the many recommendations I found on the Innerweb for handling iTunes Library moves, and it started copying duplicates of all my music files into the NFS share.

The worst part came when I told iTunes that it could manage my music library.  The Web and iTunes Help told me that with this option enabled, iTunes would keep my library organised according to artist and album detail.  Since my library was already quite structured I thought there would be little for it to do.  Unfortunately, iTunes didn’t think that my library was quite structured enough.  It started unwinding things like the “Compilations” directory — a directory it had created itself — into strict directories for the artists on a compilation.  I don’t have many compilation albums, but it was making an an absolute mess of a couple of albums that have songs by different artists.

So why was this a problem?  The problem happens because my library is shared between a number of applications — iTunes itself, ccxstream for sharing music to the XBMC-running XBoxes, and firefly (formerly mt-daapd) for streaming to DAAP-capable players (like Amarok and XBMC — I was looking at using DAAP as an alternative to running ccxstream).

So now I have a broken library.  All because I wanted to change /data/music to /lvs/music.  Why can’t iTunes manage that?  Why is it that iTunes feels it has to dereference the link to the music location, and if it thinks nothing has changed, not update anything?  As part of trying to get this working I had let iTunes move my library back off NFS to local disk, and I renamed the folder to try and hide it from iTunes as part of telling it that I had moved it.  When I went into iTunes preferences, the library location showed the name that I had renamed the folder to!

iTunes is obviously smarter than I am.  I should have not bothered trying to organise my storage the way I want to — once I let iTunes touch my network, I should have just let iTunes have its way.  So recommend all the iTunes fan-kids who posted some of the recommendations I read.

I’m much less keen to throw down on an iPod touch now, or even the Apple TV I was talking Susan into.  Maybe I’ll feel different after I’ve worked out how to put my library back together… or better still, worked out a way for me to manage music on my (and Susan’s) iPods without iTunes (my previous efforts there haven’t been successful).

We do not appreciate your insistence on devaluing your hard-won programming with the following:

* distracting and idiotic lower-thirds (believe us, in the middle of House we don’t need to be reminded when Neighbours is on.  If you intended to distract us from the programme, you succeeded — but we changed channel.);

* accelerated rolling or compression of credits to make room for promotion reels, destroying the readability of the credits (some respect for the folks that make the programs you use to make your money, please.  If you can’t take 30 seconds out to thank the folks that produce your programming, YOU’RE IN THE WRONG BUSINESS.);

* worse still, the replacement of the original production credits with your own high-speed microscopic version (see above.  Get a clue.);

* obliteration of theme music with continuous voice-overs that start from the first frame of credits until the last (in fact some of us can remember when a voice-over was exactly that, and you could still hear the original track);

* utter disregard for timeslots and scheduling, either by deliberate obfuscation (telling the guides that 8:30 is the start time, while the real start time is ten minutes later) or programming overruns (also likely deliberate.  If I didn’t choose to watch Big Brother, I’m hardly going to stay and watch it while you make the program after it late.);

* station logos that add nothing to the security or traceability of material (“This capture of Sea Patrol, Sir, from the logo we think it was captured by an AUSTRALIAN from a broadcast from Channel NINE!”).

Call it whatever you like: fast-tracked, streamed, straight-off-the-satellite, we don’t care.  Treating us like idiots is what’s driving us to torrents, not episode lag!  For many of us who value the experience of being entertained and not that of being marketed-at, we are very likely to WATCH THE TORRENT and then BUY THE BOX-SET DVD.

The sooner you reposition yourselves back to being useful and valuable components in the entertainment supply-chain, instead of annoying and costly middle-men, the better -off we all will be.

Some hints at how this might occur:

* Ditch the lower-thirds.  

* Respect your content producers, even those who are employed by your network.

* Setting your programming schedule is like making millions of appointments with your viewers.  KEEP YOUR APPOINTMENTS!

* Remove the logos.  I’ll admit that they’re less intrusive here than in some locations (or on cable) but they’re more harm than good.

Yours was an honourable industry that has been tarnished by overseas influence and corporate greed.  Only you have the power to reverse the trend and make Australian TV great again.

It’s probably fair to say that success has been limited.

The problem arises because Cisco implemented a PoE mechanism prior to the IEEE 802.3af standard being ratified.  This method, not surprisingly, is completely incompatible with IEEE 802.3af.  It’s generally referred to as “Cisco pre-standard PoE” — Cisco gear that does PoE usually supports both their pre-standard and IEEE 802.3af.

(When I come across something like this I wonder if a lot of the cost of Cisco gear is soaked up by the added complexity of having to support both the “system they designed and rushed into the market to try and preempt the standards process” as well as the “system the industry agreed on and ratified because the right way is not always the Cisco way”.  But I digress.)

Thanks to some information like this, I found out how an 802.3af-compliant switch senses that it should supply power to the wire.  It’s quite simple: the switch looks for a “signature resistor” across the cable (follow the linky-trail for more info).  This meant it was quite easy to convince my switch that it should apply voltage.  (Susan was a little disturbed by how excited I got about making a little green light come on.)

Actually getting power into the phone is a different matter.  The page I link to above has about five different versions of fact and fiction when it comes to running the Cisco phones off 802.3af PoE gear, but it misses out on one critical piece of information (or at least it will until I get an ID there and update it).

The IEEE 802.3af standard defines the partners in a PoE transaction as the Power Source Equipment (PSE) and the Powered Device (PD).  The PSE can be either a 802.3af-compliant switch, or some kind of intervening device like a PoE midspan or injector.  The standard also defines two “modes” of PoE: Type A, where the power is supplied over the same cable pairs as the Ethernet signal; and Type B, where the power is supplied over the spare pairs in the cable.

The voip-info.org page says that with a resistor and a crossover-wired cable, you can  run a pre-standard Cisco device off an 802.3af PSE.  This is only partly true, because of a very important little piece of info that’s only alluded to on the page.  The critical info is this: while an 802.3af PD should be able to work on either Type A or Type B (as you don’t know what kind of PoE source you’re going to be connected to), a PSE can be Type A or Type B.  This is particularly important when it comes to the pre-standard Cisco phones, as they can only work as a Type B-style device — the Cisco pre-standard had no way to receive power over the data pairs.

This is why folks have success running Cisco phones off midspans and injectors — because they are 802.3af Type B devices.  Type B is used when you are injecting power along the cable run (i.e. without access to the Ethernet PHY).  They then crow on about how they got their Cisco phone working with 802.3af — as always, the devil is in the detail.  In the case of trying to run Cisco phones off 802.3af Type A devices like switches, you are left with the problem of extracting the 48V out of the data pairs without breaking the Ethernet link to the end device.  Not simple.

The page above lists one switch that appears to work with the crossover-cable trick: the Netgear FS726TP.  Knowing what I know about 802.3af now, it would seem that Netgear decided to make their switch a Type B PSE instead of a Type A.  Is it wrong?  Well no, but some folk may be surprised why they don’t get power over some cables when a different switch works fine.

The good thing about the Cisco pre-standard (if there can be a good side to it) is that it should be quite easy to rig up a DIY injector using the original power supply.  Since the phone expects power over the spare pairs, there’s no need to use an adaptor to split the cable out again at the phone end.  A DIY midspan using a single 48V PSU would reduce the losses in running a number of separate power bricks too.

So if there are budding Cisco PoE hackers out there, be aware of the need to know a little bit more about your 802.3af switch than what the manufacturer says on the glossy brochure.

PS: While researching this, I came across a forum comment from someone who said that one of their pet hates was people referring to “Power over Ethernet” when it should be “Power over Cat5″.  Well, one of my pet hates is over-generalisation.  There can be no argument that the way that power is delivered in PoE is specific to the Ethernet wiring of Cat5-style cable.  Power over Token-Ring, if such a thing existed, could not be the same as PoE because different pairs of wires are used.  Likewise Power over ISDN U-Bus, Power over POTS, whatever.  My advice to Mr “Power over Cat5″: keep your generalisations to yourself, if you please!

It doesn’t have a radio, and doesn’t (as far as I know) have sound recording capability — these are a couple of features that many folks find important in a portable audio device.  Also, just because I was out of touch with my original estimate of the price, doesn’t mean that it isn’t overpriced.

16GB of storage, while impressive in some ways, is miniscule for what could be considered, thanks to the size of its screen, Apple’s flagship portable video display device.  A decent amount of storage, such as those now offered on the iPod classic, is going to be needed for a lot of people to take this seriously in comparison to something like a Creative Zen Vision or Archos unit.

Some might say the biggest criticism is the fact that, like the iPhone, there’s limited potential for third-party expansion.  Apple is a visionary company, but they can’t think of everything in advance and to not allow (or make it hard for) third-party applications to be delivered on these devices shows a distinct lack-of-vision.

Maybe I have cooled off on it, but I’m a little less keen on shelling out an AU$549 lump of money now.  Maybe Apple’s early announcement was a bad thing — they might have got a lot of impulse buyers just drop the money on it and then see the negatives, instead of (like me) having some time to think about it before being able to spend the dough.  Not that I think the iPod touch will be a failure, but given the US$200 price drop on the iPhone within six months of release I think I’ll hang onto the trusty-old third gen iPod a bit longer.

I don’t know the details of it, because it was something I only did occasionally to show that it still worked.  Now it’s stopped working, presumably after a Mac OS X update or other.  When I try and log on with an LDAP user, I get the wobbling password box.  That’s it.  The system logs tell little on the Mac, but on the LDAP server I get an error message about a failed SASL bind.

I’ve only ever set up SASL enough to support IMAPd, and even then it’s just talking to LDAP to do the work.  I use LDAP to store passwords, and for my purposes that’s always worked.  It used to work on the Mac too, but I can’t get him to stop trying to do a SASL bind to LDAP.

At about the same time as this, I was playing with a Jabber bot.  I read the instructions, configured appropriately, and it completely failed to function — its logon to the Jabber server was rejected.  Wireshark to the rescue — it was trying to use SASL to log on the the Jabber server.  Sure enough, my Jabber server was advertising SASL authmechs.  I removed the SASL settings (well, just the available authmechs) and the bot was able to log on.

So I started thinking if these issues were the kick I needed to set up a proper SASL and Kerberos system.  My ideal would be to get saslauthd to provide authentication service without having to go all the way to GSSAPI/Kerberos, something that should be possible…  except we’re talking about security systems here, so it seems that “The Right Way” is the only way.

The OpenLDAP documentation doesn’t discuss the SASL mechs PLAIN and LOGIN, since in their opinion they’re no different from LDAP simple bind.  Be that as it may, it would be nice to know how to do it!

DIGEST-MD5 is next, but the way it works you have to store user passwords in clear-text in LDAP (yes, clear-text passwords!) or use SASLDB2 to store passwords.  The former is unattractive, since I’m not so confident in getting an LDAP ACL right that would protect the password field from undesirable reading while still allowing it to be used, and the latter means I’d have to move everything to SASL auth unless I want to have password synchronisation problems (the very thing that moving everything into LDAP was meant to avoid).

Next comes Kerberos…  If I’m doing a heap of work to cut things to the DIGEST-MD5 SASL mech, might as well go all the way to GSSAPI, right?  That means more work, and again possible password sync issues between the Kerberos DB and those things still getting their passy from LDAP (although it looks as though using SASL you can tell OpenLDAP to consult Kerberos for password validation, so things using LDAP for password checking would actually get handled by Kerberos anyway).

One thing I thought to try was to rebuild OpenLDAP without SASL support — I’ve got a nasty feeling that since the last time the LDAP login worked on the Mac, I added “sasl” to the USE flags on the server.  Being built with SASL support means that slapd is offering it, even if it’s not set up (an ldapsearch for supportedSASLmechs verified this), and the Mac is seeing SASL auth advertised by the LDAP server and demanding to use it…  While a good theory, it’s not the problem.  The only difference in the log now is that there’s no message complaining about a failed SASL connection.

So after all that waffle about SASL, it looks like there’s something else happening. Likely something to do with the strange posixGroup entries it’s looking for called “ffffeeee-dddd-cccc-bbbb-aaaa-0000003c” and so on…

Back to the drawing board.

Via voip-info.org I found that a later release of the Cisco SIP firmware is in the wild, apparently with Cisco’s blessing.  I have a phone that I had originally set up in the bedroom, but it hasn’t been plugged in for a while, so I decided to drop the new firmware on it.

Pulled the files, put them on the server, changed the config files to point phone to the new version, then powered up the phone.  The loader upgraded, but then would not get an IP address from DHCP.  Argh.  Because it was in the loader (and not in full firmware) I had no way to manually set IP.  I traced the network, and it looked like the phone was simply ignoring the DHCPOFFER.

I straight away put the TFTP config back to the previous firmware version — as it seemed like the new firmware was bad, I didn’t want to risk the other phones in the house pulling the new version.

Of course I thought that there might have been a network problem, so I took advantage of the port-mirroring capability in my Linksys switch…  As far as it could tell, the DHCP response was definitely making it out to the phone.

I spent way too much time looking at DHCP options, the config of the DHCP server, and my switch setup (I tried VLAN tweaks and duplex settings in the switch).  A phone reset (holding down # while you power the phone up) didn’t help either.

I eventually found a reference to a way to do a full factory reset of the phone — after holding down # at power-up, press “123456789*0#”.  Knowing that this is what bricked my 7970, I was quite nervous about doing it.  But, since the thing ws effectively bricked already, I gave it a run.  The phone asked if I wanted to keep config, and I told it not to.  It then did a reset, the Universal Loader started, and successfully got a DHCP lease.  Because I had put the config on the TFTP server back to the previous firmware level, it reloaded the old level and before long I was back where I started.

Being a glutton for punishment, but more importantly because I knew how I could get out of the problem if it happened again, I switched the files on the TFTP server back to the new version and tried again.  This time it went off perfectly, and the new code went up without a hassle.  I went around and reset the other 7960s in the house, and they too went fine.

The only reason I can come up with for why the problem happened is that my DHCP server has changed since the last time that phone was on the network.  It’s possible that the loader kept some record of the previous lease; even though it was getting a proper DHCPOFFER from DHCP, the fact that it was a different IP or a different DHCP server meant that it threw away the offer.  Maybe.

Spurred on by my success in rescuing that 7960 from brickness, I decided to have yet another go at making the 7970 work.  Some time ago (before there was SIP firmware for the 7970) I bought it and never really got it working with Asterisk’s SCCP support…  Then when the SIP firmware came out I managed to obtain it, but in the process of trying to get it loaded I did a factory reset, which on the 7970 actually deletes the phone’s existing firmware load.  The SIP firmware would not load, because of a TFTP error during the download.  I’d removed the only functional firmware from the thing with the factory reset, so it was totally dead.  An ex-phone.

In the earlier days of the 7960, there was an issue that some older firmwares could not be upgraded directly to newer firmwares.  This issue affects the 7970 as well, and seemed to be the problem I was having with mine in getting the SIP firmware on.  The solution is to progressively load firmware versions to step the phone up to the latest version, but with no valid way to obtain firmwares from Cisco (and a philosophical objection to paying for the privilege, even if they’d sell it to me) I was stuck with a brick.

If only there was a web site around the Internet that had the files available for download…  And if only there was a search engine that could find those files on that server for me…  :)  I found a firmware that the phone would accept, and which let me go up to the SIP firmware.  After some mucking around with the XML config file, I finally — more than two years after I bought it — have my 7970 online to Asterisk!

(one of the line names is obscured, just because)

I’ve even got some of the XML app stuff working on the 7970 (as it’s subtly different from the 7960, of course) but I’ll put that in another post.  I might even post some more screenshots!

I left for a short trip to Adelaide last night.  I phoned home from the airport, and Nicholas didn’t want to talk to me… but Chelsea did.  :)

Susan phoned this morning to report that Chelsea had been ill all night.  A hurried trip to the vet…  Examinations and X-rays revealed that Chelsea had bone fragments lodged in her bowel.  Minor surgery.  Apparently while she was under, more fragments were found.  Minor surgery suddenly becomes major.

She made it through the surgery, but died later in the afternoon.

How can the light that burned so brightly / Suddenly burn so pale?

We will miss you so much, Chelsea.

I like it.  Actually, I don’t think like is the right word.  It’s… remarkable?  Fascinating?  Incredible?  I’m still looking for a way to describe it.

First thing to do is to get the vanity searches of current and past residences, if only to give you the “eye-in-the-sky” reality check (looking at someone else’s backyard doesn’t give you the twinge; it’s only when you look at your own backyard that it hits home).

Once you’ve got that out of your system though, the fun factor kicks in.  For example, I’ve spent the last few hours doing a virtual tour of some of the places I’ve been (most of which I’ll likely never return to in the real world).  And the travel directions feature is a nice addition to extend that fun factor a little longer.

Will I use it every day?  No.  But there’s no doubt that its impressive.

I was getting these errors in my log, at a fairly cracking pace:

  jabberd/sm: mysql: sql select failed: Table ‘jabberd2.status’ doesn’t exist

I’d recently upgraded jabberd2, and the config now includes a “status” module.  It would seem that this new module wants a table in the database, but the Jabberd people didn’t get around to updating the database definition!  Oops!

No worries…  If you want to use this module, one way to avoid the errors is to override the database definition for that module.  Add the following to sm.xml, in the <storage> section:

  <driver type=’status’>db</driver>

You could use sqlite as well…  Just anything that doesn’t require the database definition up-front.  Whichever you use, make sure you have an appropriate configuration section in the <storage> section of sm.xml (like you had to create a MySQL section) that defines where the database will live.

Stop and start jabberd2, and the “status” module will write to the alternative database instead of MySQL.

Of course, now we should be able to see what’s in the table, and create a MySQL table to suit, and remove the <driver> line from sm.xml.  But I’d rather wait for upstream to get it fixed.   That way I don’t risk creating the table incorrectly, with the wrong data types, or field sizes, or whatever.

If that all sounds like hard work, I guess you could also remove the “status” module from the session manager config.  Don’t know what “status” does, but if it’s a new addition that we’ve done without then we won’t miss it a bit longer.  Problem would arise if it provides an existing function that used to be contained in a different module.

Just being a Linux or Open Source user doesn’t make one a part of the community — a point that has been lost on me in recent years as I’ve been unable to participate in things like Redbook residencies and mailing lists that were my (albeit small) contributions to community.

So, in response to my web traffic and comments on the blog, I’ve decided to start up a site to release and track the little utilities I write (and write about) that help me run my systems.  If they’re useful to me, they might be useful to you — and together we might be able to make them better for both of us.

The site will be here.

First item to be released will be my little hacked LDAP Caller-ID name lookup script for FreePBX (just please stand by while I work out a nice way to place a file for downloading, until then you can view it in the SVN).  After that, I plan on creating a Gentoo Portage overlay (or at least copies of my ebuilds) for folks who are interested in the same custom versions of things that I am (like DHCP servers with LDAP backends).

There are sure to be teething problems, but please place feedback either in comments here or on the site itself.

To those who gave me the nudge to get this done, thank you!

PS: This is really part-time, and so I will be keeping it as low-maintenance as possible.  Since I have no time to monitor Wiki spam, the Trac has ‘anonymous’ wiki editing disabled right now.  When I can, and if there’s a need, I’ll work out a way for users to register and make updates.

For the first two months of the year, not very much.  Someone who searched for “mediawatch today tonight knife flight” found me in January, and in February someone (hopefully) found my experience with setting up S/Key authentication in OpenSSH useful.

March was interesting.  We went from on average 7 searches hitting us to, in March, 42 distinct keyphrases.  There seem to be as many people as upset as I am about Linksys and the IE-only web management interface on the SRW2024, with a number of searches finding that, but the standout would be people looking for Caller ID sources in FreePBX.  Again, I hope that folk found my post at least a little useful — this is the kind of feedback that makes me think I need to make these things I write available.

April saw another huge jump: 93 distinct keyphrases — on average three a day (if that’s a meaningful distillation).  Trixbox was the most common search word, and Caller ID lookup was again popular.  Problems with kwlan also brought people here, which is surprising as I don’t recall using kwlan let along blogging about it…  Top search phrase for the month though was “xbmc ethernet into wall bypass router”; not sure what this person was trying to do, but it sounds tough.

May saw 102 keyphrases, and again permutations of “ldap caller id trixbox” were plentiful.  Top two were “trixbox ldap” and “freepbx ldap” –  I really need to get some material out there.

To round out the review, I’d like to mention some of the more unusual searches that found their way here:

* “places to get married qld dicki beach” — good luck with the nuptials!

* “how does cifs work hummer” — err, okay… every Hummer needs a CIFS server, I’m sure

* “angiogram recovery time” — I hope it went well…

* “vic s blog” — I see Karl Malden: “Ask for it by NAME.” (gee, does that show my age?)

* “wedding transport services sunshine coast hummer” — the Hummer is back!

* “bacula volume frustrated” — yep, Bacula can be frustrating; stick with it though, it’s good gear!

* “bluetooth in hsv grange” — I have NO idea how that got here!

* “wedding venues dickie beach” — not sure how these wedding nights are going to run if these folks can’t spell Dicky…  :)

* “laporte telephone voicemail system virus” — at a guess, someone’s looking for something Leo might have said about a virus that spreads through voicemail…  only the Great GoogleBot could find a way to send that to me!

One thing is clear to me though — I spend a bit too much time blathering on about things I’d like to do, or haven’t gotten round to doing, or whatever…  I know when I’m looking for stuff I hate getting sent to some waffly blog that tells me nothing.  So from now on there’ll be more content, or at least references to where I keep more tech stuff.  Who knows, even a separate tech blog.

Thanks for reading, and if you’re a visitor who’s found this site via the wonders of the modern search engine: perhaps it’s exercising Zen Navigation (As described by Dirk Gently in Douglas Adams’ “The Long Dark Tea-Time Of The Soul”)… the art of navigating by finding someone who looks like they know where they’re going, and following them…  in doing so, you rarely end up where you wanted to go, but often end up where you have to be!

I had played with the MythTV frontend for XBMC a few times, but each time there seemed to be something wrong with it.  Either dodgy config, Samba problems, or insufficient bandwidth over the wireless.  So I figured that conversion to MythTV and plugins like MythVideo and MythMusic would be in my future.

Then, I realised something wonderful.  All the files I was creating on MythTV had an extension “.mpg”.  On a whim, I set up ccxstream (the XBMSP streaming server) on the MythTV box (sometimes it’s good to run Gentoo for this stuff) and pointed my XBMC at it.

It played.  IT PLAYED!!!  XBMC saw that it was MPEG2 (and good) and Just Played It.

Now I’m sure many readers are looking for a virtual wet fish to poke me in the eye with.  I don’t know why I never thought to try it before, especially after I set the “Transcode” option on all my recordings in MythTV.  But with Mythrename set up to give the recordings sensible filenames, I’m happy — most importantly so is Susan, who took the opportunity to watch some of the episodes of her soap using the XBox.

There still is the issue of XBMC freezing occasionally (nothing to do with the MPEG2 stuff from MythTV as it does it with MP4s too, probably something thermal), and the nagging question mark over network bandwidth, but I get the feeling that my mythical killer home entertainment platform is within my grasp.

I found some info on the ‘net about how to get a Netgear DG834 to just be a router rather than being a total PITA firewall.  The link that got me going is here.  I thought I had done this on my previous try with the router, but perhaps it needed the firmware upgrade I did this time out.

So with these firewall rules in place, I was up and away.  There was a problem with connecting to MSN via the Jabber PyMSNt transport (yes, I MSN, but only because my family does, and I can’t get them off it because their friends MSN…  see why Open Communication Day is bound to fail?  But I digress.), but I had changed other things (like doing an emerge -uav world on the Jabber box that pulled in a glibc update, which I’ve found always results in a few days of fighting little problems).  A week later I still couldn’t MSN, and after re-emerging everything I could think of on the Jabber box I thought “maybe it’s network-related”.  Sure enough, exact same problem using Kopete to connect direct to MSN.

The Netgear DG834 is one of the many devices using an embedded Linux distribution (yay Netgear), and it provides a hacky way of getting a shell prompt.  Getting this shell prompt and running a few iptables commands told me what the problem seemed to be — a rule to redirect all MSN and AIM/ICQ traffic to a local port on the router.  The router comes with a bunch of Trend Micro security crap pre-installed, for which you have to subscribe to keep up-to-date (boo Netgear).  It seems that without this security rubbish activated, there is no process active to be able to handle the redirected traffic.

I think it’s a bug that they give you an option to disable their security rubbish, but leave firewall rules in place that result in traffic being killed if the stuff isn’t running.  I could have worked around it though: using the shell I could have just deleted the rule.  Then I’d have a problem of having to do that every time the router decided to update the rules.  Could I hack the firmware and remove the rules from the image?  Possible, but the stuff is probably signed and a hacked one might not run.  Maybe one of the free/open replacement router firmwares will run on the DG834?  Haven’t looked at that.  Of course, given that I think it’s a bug, I could also file a support request with Netgear too…

I had originally titled this “Netgear: strike three, you’re out”, but given that I have a few options available I can give the thing one more chance.  But it’ll be another day; I’ve stuffed around with the thing too long this time out.  There is a bit to like about the Netgear, so I admit to being tempted to persist with it; the built-in switch lets me think about upgrading the Linksys WRT (running OpenWRT) that I have as DMZ router and wireless access point to something that is 802.11n but won’t be able to do the DMZ routing.

For now though it’s back to the trusty old Alcatel Speedtouch, that I’ve had since my first ever ADSL service.

I subbed in a slightly more expensive motherboard, and when it came to the graphics card the one they offered had less RAM than I’d have liked (having had the black-window problem in Compiz/Beryl) but the next one up they had was nearly double the price.  I didn’t even think about the chipset while I was there.

In fact I didn’t even think about it until I got home, and sure enough found that the card was an ATI.  Nevertheless, I threw it all together, thinking “oh well, surely the driver support must have improved since I last used an ATI card…”

W R O N G.

The new kit went in underneath my existing Kubuntu Feisty install, which had Compiz running acceptably on a moderately-recent Nvidia card.  First thing I had trouble with was the right options to get Compiz going again: I went straight for the ATI binary driver, but then backed out to the Xorg Radeon driver because MythTV wouldn’t  work right (the ATI driver doesn’t support video overlays in acceleration mode, or something).  I ended up having to ditch Compiz altogether because it was just really really unstable.

So I’d lost my eye candy, but had TV.  Then, I wanted to look at FlightGear (the FOSS flight simulator)…  I was getting about 3spf (yes, that’s seconds per frame, not frames per second) in the game…  Grr.  The Xorg ATI driver has no 3D acceleration support!

By this time, I had blown about a week of whatever spare-time I might have had trying to get the ATI card working properly — and it still wasn’t.  Not only that, but it couldn’t: the configs needed to support the apps I want to run are mutually-exclusive.  So I spat the dummy, and went to the local PC-bits shop and bought an Nvidia card.

Threw it in there, chucked the Nvidia binary driver at it, and it Just Worked.  MythTV: perfect.  FlightGear: brilliant.

Compiz is a different story though, as it still is dodgy on the new Nvidia card[1].  Might be a bug that’s unrelated to the graphics driver — odd that it only showed up when I put the ATI card in though.

Anyway, I’ve got an ATI X550 based PCIe video card here, going cheap…  :)

[1] The first version of this blog post was lost to a Compiz-induced X crash…  While I was typing, I thought “hmm, now that I’ve got the Nvidia card working and things are stable, I should try Compiz again, so that I can blog it”.  Hint: when trying something that has killed X in the past, don’t do so with unsaved work in your desktop…

The device has a cooling arrangement which seems to be comprised of a bunch of those incredibly annoying 1″ microfans (usually seen in low profile “pizza-box” style rackmount servers).  The unit makes a heck of a noise while the cooling is running — I can hear it from two rooms away, even over my tinnitus.  :(

I had seen a bunch of comments about noise from the network version of the 1TB device, but I mistakenly thought that the non-network device would be cooled differently.

Sigh.

I think the real problem comes from Mac OS X not allowing the drives to become inactive.  In Activity Monitor, there is a constant 3-4 disk writes per second which (if not activity caused by the monitor itself) might be keeping all the disks active.

Also, there’s something called “WDDrvSvc” that’s eating a few percent of CPU; I would think that’s just the service that keeps the pretty lights on the front up-to-date, but 3-5% of CPU is a bit much to pay…  Kill it, and it gets restarted immediately…  Sigh again…

I haven’t Googled anything about it yet, but if nothing turns up I can go back to my original plan of attaching the drive to the Slug and see if that doesn’t keep it busy.  It’s also possible that running it in RAID-1 mode instead of RAID-0 keeps everything busy.  I have verified that a single-drive MyBook does spin down when attached to the Slug.

Update: WD has a firmware update that changes the fan operation, but from the description in their knowledge-base of the changes it sounds like my unit already had the newest firmware.  There are also some hacks around; folks have replaced what seems to be the poor-performing fan WD used as original equipment (I was wrong about them using 1″ microfans) with a larger, quieter fan with greater airflow.  Others have flipped the fan around, because WD has the fan blowing air into the case from outside and it seems to be better swapping the fan to blow air out of the case.  Some wisehats have even decided that the case has little airflow provision, and have taken to it with their Dremel tool…

Tonight, I went into a store — not even a computer products store, mind, just a general office goods store — and purchased one terabyte of disk in a package that runs off an AC adaptor and I can fit on the palm of my hand.  And, the cost was considerably less than $1/GB.  The unit was on the shelf, with another of the same type — no special order or promotion, just normal store stock.

The madness doesn’t end there: for $50 more I could have gone across the road to a different store and got the network-attached version.  That store had four of those units on the shelf.

The unit I bought is the WD MyBook Premium Edition II.  I looked for a while at the network version, (MyBook World II) but although the embedded OS is Linux-based and a throwaway line in a Wikipedia entry mentions an Open firmware can be placed on it, I could find no evidence that this is the case.  The unit only provides Windows shares, and that’s not so useful to me.

I configured my device in RAID-1 mode, so the 1TB (I should say 1 trillion bytes, as usable base-2 capacity was about 932GB) becomes about 465GB.  I’m attaching it to the Power Mac using Firewire-800, which should make for a nice storage area for video editing and the like.

My main machine that ran Windows on an almost-daily basis was the Sony laptop, and it’s running Kubuntu Feisty nowadays (except on rare occasions when I boot XP).  Susan’s laptop is the holdout (running XP Home because of a crappy wireless card that I can’t get working with ndiswrapper), but she does no file server access — she prints occasionally, but converting her printer connections to IPP will see that off.

Having said all that though, I just know I won’t remove Samba.  It’s just too…  I don’t know…  It’s like that shifting-spanner at the bottom of the toolbox — your dad always taught you not to use shifters because they burr the nuts, but there’ll always be that one Imperial bolt somewhere that your Metric ring-spanners won’t fit…  Or the overtightened nut that someone else already burred with their shifting spanner…  Or the days when you’re just too lazy to take all your separate spanners with you.

Probably as far as I’ll go will be to remove Samba from automatically starting on the server.  For times when I boot the laptop to Windows and need something I’ve kept on the server, I can start Samba manually.

A home network that’s free of SMB/CIFS…  Yay!

The newer version of the software I’m using has RSS support, so if you want to stay up to date with photos of Nicholas (and who wouldn’t!) you’ll be able to pick up a feed and go.

Explanatory text (or, why did it move): I’m distancing my personal stuff from the company domain, something I’ve wanted to do for ages but just never got around to.  I got a bit of a prod from someone who wanted to look at the gallery but found it inaccessible, so I did the deed.  In time, the move will apply to pretty-much everything veejoe.com.au, including e-mail addresses…

Of course some anti-Francophiles just have to mention that they couldn’t break the overall speed record for a train, which still stands at 581 kph, set by a Japanese mag-lev train.  For me though, the fact that good ol’ steel-wheel-on-steel-rail managed to get within seven kilometres per hour of a mag-lev train says a lot about the French technology.

It’s probably fair to say though that the mag-lev rig probably has a better chance of entering service at speeds approaching the record, while this TGV run was purely a record-setter — a special train set up for the run, with fewer coaches, larger wheels, bigger motors and higher AC tension on the overhead.

Riding TGV at speed is still one of my life goals.  The closest I’ve come so far is Thalys from Amsterdam to Brussels — it runs the original TGV stock, but in Holland it doesn’t run to top speed because it’s only on standard track (not the high speed purpose-built tracks built for TGV in France).

There’s some video of the latest record floating around the ‘Net (a search for “TGV rail speed record” on YouTube will hook you up) that is just astonishing.  People standing on rail bridges while the train flashes by underneath them.  A pan following the train by a camera that must have been a mile away to be able to keep up with it.  Some footage from an aircraft matching speed — something impressive in itself, a plane flying at that speed while low enough to be able to keep a ground vehicle in a fairly close camera shot!

Okay, obviously I’m a train nut… but this is COOL!  Vive la TGV!

The wedding photographer was introduced by the M.C., and proceeded to announce that he was going to do “something special” for everyone to remember the day.  He directed everyone’s attention to a screen, and started a slideshow of the photographs taken so far in the day, set to music.

I had noticed during the ceremony that the photographers were using DSLRs (Nikon mostly, if you’re into the details).  In case you still had doubts, it looks like film really is dead!

I was watching the people in the room while the show was playing.  There were tears in some eyes.  It was amazing to see how a technology shift could add a new dimension to something, the way that simple five-minute slideshow did to that wedding dinner.  I wonder if it was through something like that, the use of digital not just to replace film but to transform the way images are used and presented, that made Kodak finally say “okay, you win.”

People who were at that wedding (and I’m sure every wedding that photographer does) will be talking about that for weeks, perhaps months.  All for a few minutes’ work in front of a computer (heck even I’ve thrown something together with iPhoto and a swag of Ken Burns Effect in about half an hour, and that was starting from scratch).  Kudos to artists like that photographer who can bring such joy and happiness to people through their art (and a splash of tech!).

To Michael and Cassandra, the very best of good wishes for your future…

Alcune volte basta un attimo per dimenticare una vita, mentre non basta una vita per dimenticare un attimo: quello che oggi vivete e che durer? tutta la vita!  Possa la realt? essere pi? bella dei sogni che vi hanno portato oggi all’altare.  Il mio augurio pi? affettuoso in questo giorno di festa.  Possa il Signore benedire la vostra unione.  Possano i vostri giorni essere pieni di amore, cura, sostegno e passione.

…which I’m told translates as “Sometimes an instant is enough to forget a life, while a life isn’t enough to forget an instant: that which you are living today and that will last your lifetimes!  May reality be more beautiful than the dreams that have brought you today to the alter.  My best wishes in this day of celebration.  May God bless you on your wedding.  May your days together be full of love, care, support and passion.”

The cutover to Trixbox changed a lot of how we worked with the phone system — and not all for the better.  Losing LDAP caller-id lookup and other features was a bit of a PITA, and honestly was a bit too much to sacrifice just to get the easy configurability of FreePBX.  I blogged about how I restored the LDAP caller-id lookup, and that the LDAP phone directory was on the way back, but I remembered one other thing that was really useful in the previous setup — voicemail.

In an office phone system, you would generally associate voicemail with an extension.  FreePBX follows that model, and makes it really easy to set up.  For a home system, however, it’s not ideal, because (at least in my setup anyway) there are a number of extensions that are all in the same ring group and are actually all for the same person (or people).

Right at the start, I configured this in FreePBX by setting up “Other (custom) device” extensions with Voicemail enabled, and disabling voicemail on all “real” extensions.  In the setup of the real extensions, I set the “mailbox” field in Device Options to be the correct voicemail number that extension would belong to, so that the Message Waiting Indicator (MWI) worked correctly.  The trouble was that getting into the voicemail system was not a one-touch deal any more, as the default FreePBX voicemail access feature codes were all set up according to the mailbox-per-extension model.  So we had to dial the Voicemail number (a non-intuitive number starting with *98 according to FreePBX) and log in with a password.

I thought about setting up a custom application (FreePBX-speak for a hand-written context in extensions.conf), but thought that there must be a way to get the existing code to work…  but after looking at it for a while, I decided that a custom app would be much easier to get working my way, and be less likely to break in a future FreePBX release.

The end result was incredibly simple.  Since I had already told FreePBX the desired mailbox for each real extension (the MWI config I mentioned earlier), I simply wrote a few lines of dialplan to extract that from MySQL and send the caller to that voicemail box!  I was expecting a lot of work, but I think it took longer for me to get the MySQL userid and password right than it did to get the code correct.

Other things I did were a rearrangement of the queue and IVR for the work phone line, and I made a bunch of system recordings for queues and voicemail.  There’s very little left for the phone system now!

In the last few weeks, AMD have released Athlon 64 X2s with higher clock speeds, and correspondingly dropped the price of the existing processors.  The prices of some of the motherboards I was eyeing have dropped slightly as well.  AMD was back on the agenda again!

Then, I saw a listing for the Athlon 64 4×4 chip, which uses Socket F rather than the Socket AM2 of the X2 Athlons — Socket F being the same socket used for the newer Opteron parts.  So now, putting a Socket F board into the desktop machine gets a look-in… but then, I could upgrade the Opteron server machine and cascade the current dual-Opteron rig to be the desktop…

Choices, choices!  Maybe I should just stick with the old laptop — dud hard disk (I need to turn the machine on its side when I power it up) but it works okay…  kinda…  Ooh, ASUS M2N32-SLI Deluxe…  shiny…

I’m using FreePBX for keeping my Asterisk box configured, and it’s working great — but there are a couple of tricks that I just haven’t quite been able to work around.  You see, FreePBX applies a few assumptions that work fine in a small office environment (like voicemail-per-extension) but don’t map to home use (one voicemail for the household).  The biggest aspect of FreePBX is that it maintains the dialplan in a set of files all of its own making — you can extend it using “custom applications” (basically a reference to a dialplan file of your own creation), but I’d be concerned about an upgrade trenching custom dialplan files…

My original LDAP caller-id lookup module implemented a new dialplan app (something like  LDAPCallerName).  I inserted it into the dialplan at the relevant place, et voila, name lookup.  But even if I could port my old code to the later revisions of Asterisk, or use the community LDAP lookup module (which was written just a little while after I wrote mine), how do I add the lookup into the dialplan?

FreePBX has a module called “Caller Name Lookup Sources”.  Out of the box, it has the types “Internal”, “ENUM”, “HTTP”, and “MySQL” (“SugarCRM” is listed as well, which I figured would be tied into the SugarCRM system that’s provided with Trixbox, but when you select it you get “not yet implemented”).  MySQL doesn’t help me, as I don’t want to transfer data out of LDAP into some other store.  ENUM is a name lookup system based on DNS, and having seen a lot of work on LDAP-backed DNS servers I thought this might be interesting… until I realised that I’d likely have to spend a heap of effort adding the DNS attributes and object types to my existing LDAP data (assuming that I could use the data in its existing structure at all).

I had disregarded the HTTP method as overly clumsy — on a single host, contacting the HTTP server to run a script to get data from the LDAP server just seemed too much overhead to me.  After I had a think about the available options though, it made a lot of sense — and half-an-hour after I decided to do it, I had a working prototype (and I consider myself very much a PHP newbie).  My script has a few nice features that tie in with the way the Caller Name Lookup module works in FreePBX, including the thing that I found was missing in the Asterisk LDAP module — the ability to drop leading zeroes from the number to be looked up, allowing you to have numbers stored in your database in international format.

Next to come will be the directory lookup CMXML app, that will allow the “External Directory” function on the Cisco phones to work again!

The new machine was not in any of the Cacti graphing, so I decided to set it up on Friday night.  net-snmp had been installed, but I must have removed it in an overly-agressive cleanup (when I transferred the desktop built onto it).  So I rebuilt net-snmp, configured and started it, added the box to Cacti, and was happy.

The following night, Susan told me that her mum had tried to phone in the morning and it rang out.  At first we figured we were actually out, but the time was wrong.  Looking in the Asterisk CDR log, I saw a call of 0:00 duration, marked as “NO ANSWER”.  Picked up the mobile and dialled home, and sure enough no ringing.  What was worse was that I could not even dial between extensions.  Hmmm.

Googling turned up little response, and neither did turning up the log level.  I could find that a curious AGI script called “dialparties.agi” was telling the dialplan that all available extensions were busy (which they certainly were not).  By the output, it looked like the script was not running correctly…  I found that there was an update to FreePBX available, and thinking that I might have tickled some bug all of a sudden I tried to install it.  The installation script produced a cryptic error, something like the following:

/usr/bin/php: libnetsnmp.so.5: cannot open shared object file: No such file or directory

Suddenly the soup got clearer!  Rebuilding net-snmp probably brought in a version upgrade; PHP was linked against net-snmp, and the library was gone.  But the whole FreePBX interface is PHP code, and it was working fine… it would seem that Apache mod_php was either rebuilt after the net-snmp build or does not link against net-snmp, because any PHP code in web pages was fine.  And sure enough, the dialparties.agi script is command-line PHP.

So rebuilding PHP fixed the problem!  Now I know some more about FreePBX — it is totally broken without a functioning command-line PHP.  PHP in turn is dependent on net-snmp… and who knows where the chain goes after that…

I probably need to put this kind of stuff on my Wiki, as it will likely catch me out again.  Maybe I need to get in the habit of running revdep-rebuild a bit more often (or compile a list of those packages that revdep-rebuild should *always* be run after).

EDIT: Zoinks!  revdep-rebuild has a list of about 50 packages that are broken…  Sigh

He loved the Lamborghinis, was in awe of the Audis, thought the Ferraris were fine, and took a liking to the Lotus.  He also thought the Nissans were neat, and took a shine to the Subarus!  While at Holden, he couldn’t take his eyes off the HSV Grange.  He climbed in a few cars, ran his hand along a few more, and even sat on a couple of scooters — I think he liked the Yamaha Majesty best!

Pop and Nicholas sat in a Peugeot 407 Coupe and were both impressed (even though Nicholas said the radio was “a bit noisy” when Pop turned it on!).  His eyes almost popped out of his head when he saw the Hummer H3, but it was on a tiny little stand that was way too crowded and he couldn’t get away from it fast enough.

There were a couple of Corvette Z06s there as well, and while Nicholas didn’t think too much of the car, he was glued to the screen showing video of the car being put through its paces by the presenters of such TV shows as Top Gear.

By lunchtime, we were all totally worn out.  Nicholas fell asleep on my shoulder as we were getting ready to leave the floor.  As worn out as I got though, I can’t wait to take him back next year!

I did occasionally get problems with repeated posts when viewing the Dudeville RSS feed in Akregator, but that hadn’t happened for a while.  Let’s see how this one goes…

Around the same time (actually while I was waiting for software updates and reboots of a bunch of systems for work) I made myself a hackergotchi.  Planet is adding it to my posts in both the HTML and RSS versions of Dudeville, and I’ll be adding it to the CSS for the blog too.  If you’re wondering what I’m talking about: hackergotchis are the little floating-head style pictures.  I think mine came up alright, considering the model…

I went looking at a couple of the generators mentioned in Soren’s article.  Freesafe looks very nice, with the ability to securely store the passphrases for a number of hosts separately.  It looks very nice and has a lot of features, but the developers have had to sacrifice the generation of the “six words” format of the one-time password.  Since you enter the OTP at a traditional no-echo prompt, the idea of entering a long OTP in hex digits was not appealing, so I kept looking.

Eventually I found one called vejotp (the similarity to my moniker is coincidental; I’m not associated with it at all!).  It remembers your passphrase and sequence number, so each time you need a OTP you just hit the “Go” button and it gives you the goods.  It only remembers the details of one server though — so if you have more than one system protected with one-time passwords, vejotp is not really for you (unless you like tapping out strong passphrases in predictive text!).

Soren’s post has more info about how he set it up on Ubuntu — for me it was as easy as adding the “skey” USE flag and re-emerging openssh, making sure that brought in the skey package if it wasn’t already there.  Then, I updated sshd_config to “ChallengeResponse yes” and “UsePAM no”, restarted sshd, and the setup was complete.  Each user would then run skeyinit to seed their OTP generation.

The nice thing is that passwordless login using SSH keys still works, so I can use the existing method of key-based login from inside the network, and OTP when I’m out-and-about.

Soren also describes a way to use OTP for mail access using IMAP and a utility called imapproxy.  Well worth the read!

The Linksys SRW2024 is a managed 10/100/1000 switch at a rock-bottom price, and since I’ve had a really good run with Linksys kit I decided to give it a go.

It has a console port on the front, but that’s only good for basic setup (IP address, etc).  The main configuration interface is via a web browser.  I got the IP address in via the serial console, and fired up Firefox to check out the web interface.

The login panel appeared, which I completed, then garbage.  The menu screen only partially loaded, bits of graphics were overlaid text, and none of the menu options worked.  Hmmm.  I started checking network connectivity — a flood ping to the switch’s IP was all good.  I opened the page in Konqueror, and got a bit further, but still the menu was non-functional and the screen was not rendering correctly.

At about this point, the unthinkable idea emerged — was there still a company in the 21st century making devices with IE-only web interfaces?  Sure enough, I went to the Windows XP virtual machine on my work laptop, and the page worked perfectly.  Aargh!

A bit of research on the web would show that early versions of the firmware on the switch worked okay with other browsers.  And, looking at the Javascript console in Firefox shows a raft of errors in the CSS files — so it seems to be sloppy coding rather than exploitation of non-portable features.

So who’s the enemy?  Linksys?  They’ve turned out a shoddy firmware that doesn’t even have valid CSS in the management web interface — what does that say about the switch code itself?  What about Microsoft, who blather on about standards but still make a browser that doesn’t care what kind of horsemeat it renders?

I’m yet to try Opera or Seamonkey — I don’t expect Seamonkey to be any better than Firefox, but Opera might give joy.  I think I can forget about Safari too, since I saw one review buy a guy who returned his SRW2024 because the management interface wouldn’t render on his Mac (a bit extreme I first thought, but if you don’t have any Windows in your shop at all, the cost of a Windows box just to manage your switch blows the cost-case a bit!).

Stop Press: I tried the interface again with Konqueror on the home laptop, running Kubuntu Edgy, and it worked!  Not completely, as the rendering is still a bit dodgy and I can’t actually configure anything, but most of the viewing screens work just fine.

I traded some credit-card-award-scheme points for a discount on accommodation at the Conrad Treasury Hotel.  A lovely old building, transformed into a high-class hotel.  The development retains the incredibly (almost vertigo-inducing) high ceilings and ornate fittings of the original building.  They even kept the knee-high door-latches!  Susan and I met friends for a dinner at the Hilton and spent a little time at the casino before calling it a night, then had a buffet breakfast in the hotel restaurant the next morning.  Very relaxing, and doubly so since we arranged a sleep-over at Nanna and Pop’s for Nicholas

For me, work has been an utter drag.  I’ll talk about that in a separate post (perhaps).  Susan is on a working adventure, though, having been accepted on a 12-month posting into their Systems Testing section.  I’m really proud of her, not only that she got the posting but that she got into gear and applied for it in the first place!  As her background is processing rather than systems, she is likely to be covering more of the payment-specific parts of their work — they have other folks who have a background in system testing that do the more system-technical stuff.  It looks like a really challenging role for her, and just the thing she needs to restore some sanity to her career.

Susan’s change means that she leaves very early to go to work, and on public transport.  So, I now have the important duty of taking Nicholas to and from day-care!  Talk about eye-opening — I’ve had one more object lesson in the incredible variety of tasks that makes up parenting.  Some mornings are better than others, and there is definitely an art to leaving him there without tears — his and mine!  He’s adapting to the new schedule well, but it’s only been one week so we’ll have to see — but I’ve no doubt he’ll be fine.

PS: I always try to come up with some witty way of closing these posts, but this time I came up dry.  Oh well, next time!

Since the servers and the network gear are currently in the study, the UPS being in the garage produced a bit of a disconnect.  It all came together though when I realised that the electricians were going to run the Cat 5 to the same place in the garage.  By relocating the network kit to the garage as well, I could achieve a bit of a reduction in heat and noise generation in the study.  The downside was bringing a stack of network cables (and to a lesser extent the power leads) back into the study…

I figured that I just needed to pop my network patch board through the wall and I’d be done.  It almost ended up being that simple, too!  A local electrical supply store provided me with some cabling grommets that I could use to neatly and professionally throw my cabling through the wall between the garage and the study.  A bit of re-terminating later and 16 Cat 5 cables (with a patch panel at each end) ran from garage to study without running through doorways!

The last issue to overcome was monitoring of the UPS.  The supplied serial cable was way too short, and even the extended-length one that came with the old rackmount UPS was never going to make it.  The new UPS has a USB interface with a special cable (I’d never seen a 10-way RJ45 in the wild before), but USB is even more affected by cable length.  I know that serial can be run over even old Cat 3 cable, but being a “smart” UPS cable there was a very good chance it wouldn’t work.  I thought about running the USB over Cat 5, and found that too had been done with little more than a couple of jacks and a soldering iron… but with warnings a-plenty about current limitations and other “dangers”, I was reluctant to get the solder-station out just yet.

Google to the rescue again!  I found a mob that produces a device that extends USB up to 50m over Cat 5 cable.  Better still, their Australian operation is right here in Brisbane!    (That’s the reason I was journeying to Brisbane’s northside recently, see other post…)  Props to LINDY Australia, your little Cat 5 USB Extender and one of my Cat 5 garage-study runs got my UPS talking to my server again!

Now all that remains is to bring the phone/VMware server back downstairs so it can join in the new-UPS fun (a task enabled by the extra Cat 5 runs the electricians pulled), and the work will be all complete.

At first I was just going to replace with an equivalent.  The old one was nearly eight years old, however, and “they don’t make ‘em like they used to”.  The rack I have is only a tiny wall-mount type (that I put castors on so it runs on the floor) and is nowhere near full depth — but virtually all the APC UPSes now are at least as deep as they are wide (the 2200VA 2RU job is about 650mm) and are way too big for my housing.

So I decided to go with a tower (floor-standing) version, and with the cost-saving from doing that I could more easily justify an upgrade to the 2200VA version.  A very helpful fellow at PowerFirm told me everything I needed to know, and toward the end of the conversation said “now, you know that these units need a 15A plug, right?”  Well of course I didn’t…

So then I rang around looking for the electrician that would do such a job in the fortnight leading up to Christmas.  I found one, and a couple of days later they arrived.  I indicated where the point was to be installed, right near an existing one in the garage.  Sparky replied “well, it’s got to be on it’s own circuit.”  Oh, crap, well of course it does, doesn’t it…  ”Where’s the power box?” he asks, and I reply “the other side of the house…”  Both our faces fall: his upon seeing a 15 minute quick fill-in job turning into possibly hours of pulling cable through sealed walls, and me upon realising that the job that I could pay for with the cash in my wallet turing into something far more expensive.

I figured I needed to make the most of the situation, so I asked the guys to pull some Cat 5 for me while they were in the walls (which leads to Vic’s Household Hacker Tip #1: Always keep 100m or so of Cat 5 lying around; you never know when you might be able to get some cabling done).

So the stage was set…  all I needed was the UPS to arrive…  Stay tuned for Part Two!

Recently I had to locate a store on the north side and, armed with the address and a fifteen-year-old knowledge of street layouts, I set off.  I did eventually find the store (more on that in another post, perhaps), but in the process got hit upside the head by a realisation that even in Brisbane the pace of change can overtake the unwary.

All I needed to have done was look at a street directory before leaving, and I would have realised that I was not headed for the Sandgate Road, Albion of my youth.  Instead, I blindly trusted my ancient knowledge would guide me safely.

To be fair to myself, it’s not like the whole of the northside has been torn up and rebuilt — the part of Sandgate Road that I was looking for just happened to be the part that got a freeway parked on top of it.  That general area of the inner-north of Brisbane got fairly well rearranged by the Inner-City Bypass.  In the end, my ancient knowledge did save me from getting completely and hopelessly lost, but the exercise ended up being a lot longer than the quick duck-out-before-work-in-light-school-holiday-traffic trip I was hoping for.

When I moved the phone system to Trixbox, it was because the previous system was stuck on an older Linux distro (a SUSE 9 that just refused to upgrade) and the Asterisk build was from-source.  I was keen to set a system that Just Worked, as close to out-of-the-box as possible, was upgradeable, and didn’t require that I do a lot of manual work to maintain.  Trixbox certainly gave that in the first instance, and came with a lot of extra function (web-based configuration, head-up display systems, a CRM facility that I was going to use for…  I dunno, but it’s really cool)…  However, I had to do a lot of mucking with it to get ISDN working, and I’ve had to give up the LDAP caller-ID integration I had set up.  Voicemail access is confusing and different to the original system (it’s probably fortunate we don’t get many messages), and because I foolishly decided that I would run VMware Server on the same box, I’ve blown the “single-use” idea.

KnoppMyth looks like it might head the same way.  It has a lot of extra function built in, like the pre-configured MythTv plugins, that normally would be a lot of extra work to integrate.  However, I really want to get this new tuner card working, which looks like I’ll have to build not only MythTV itself but probably the kernel and the DVB support utils from source.  That blows the “stock-built” maintainability of KnoppMyth.  Also, since KnoppMyth uses the Debian repositories, APT reports that there’s nearly 400 packages that are available for updating — but I can’t find anything in KnoppMyth doco that says whether it’s safe to do an “apt-get upgrade” on it.

I’m wondering if I might be better off sticking to my distro-of-choice (still Gentoo currently, although running servers on Gentoo can be a bit of a pain and I might be looking to change) and just do the grunt work myself…  I’m arguably competent enough to handle building things myself, so maybe I don’t need the handholding that these built-to-purpose distros provide.

(Thinking outside the box a bit, I wonder if the VMware folk might have to start being a bit careful about how they “market” VMware Appliances, for the same reason.  Folks might download a heap of appliances for different purposes, then some time down the track realise that what VMware has let them do is make dozens of different systems that might not even all be the same OS (there are FreeBSD and OpenSolaris appliances on VMTN as well as Linux) that all need to be maintained in different ways.)

Something to ponder while I wait for my next “emerge -av world” to run…  :)

She got rolled by a couple of kids on her way from the railway station to home (she could probably have seen her front door from the spot).  While they didn’t actually attack her so much (rather they grabbed her bag, knocking her for six in the process), she did come out of it a bit banged up and certainly shaken by the experience.

Those unfamiliar with the surroundings would jump to the conclusion that she should have stuck to the better-lit pathways rather than the short-cut she’s taken in safety for the last 10 years or more.  Halali-like, they’d say “she was asking for trouble”.  Not likely.  Like any kind of attack, who knows how many times in the past the perpetrators were ready to strike and were warned or scared off by something at the last minute.  More likely, this was something random, a wrong-place-at-the-wrong-time kind of thing.  Besides, in this case “better-lit” is no more than a dodgy streetlamp at 100-metre intervals.

While it doesn’t help the situation, I’m bloody angry about it.  I’m angry about a company that thinks it’s okay for a female staff-member in her sixties to make her own way home at midnight.  I’m angry at an establishment that sends its police and army overseas to fight in multiple theatres in a fictitious “war on terror” when there is very real terror in our own suburbs.  And, despite what I wrote earlier, I’m angry that she walked in the dark…

Most of all, I’m angry at a son who doesn’t do anything about his mother walking in the dark.  Could I have done anything?  No, but still…

Here’s hoping the little bastards get everything they deserve — and I’m not talking about our piss-weak civil-libertarian-kowtowing give-them-every-chance-under-the-sun-to-rehabilitate-while-they-reoffend-at-taxpayers-expense judicial system…  No, I’m talking about karma.  Good luck in the afterlife guys.

The first thing I thought of was a problem caused by daylight savings (the state of Queensland doesn’t observe DST, and those Aussie states that do transitioned a week or so ago).  But where?  And why now all of a sudden — I’ve been running Cacti for a few years now.

Strangely, the entries in the cacti log file were timed correctly, except for the entry made by the poller job at completion.  So it seemed that it was just the poller job — which runs from cron — that had a dodgy setting.  I checked for TZ environment variables in apache and cron, and found nothing.

Then I thought of running phpinfo.  If you’ve not done PHP work before, phpinfo() is a built-in function in PHP that dumps a very thorough list of current configuration settings and environment details.  Running phpinfo() through the web server showed that there is a default timezone setting in PHP that was defaulting to “Australia/ACT”!  Changing this to “Australia/Brisbane” in php.ini fixed the PHP code running through Apache, but didn’t change the graphs…  On Gentoo, there is a separate php.ini for CLI PHP — making the same change there fixed the problem.

So why was this not an issue last DST change?  I’ve upgraded to PHP 5 during the last year, and looking through the php.ini from the old PHP 4 installation showed no timezone setting.  So perhaps PHP 4 picked up timezone from the shell environment…

So I now have correct graphs in Cacti, and posts in this blog (Polarblog is PHP-based) show up in correct local time again.  One more for the good guys!

The flat-pack box had a couple of boot-prints on it — always a good sign, but seemingly par-for-the-course.  Opening the box displayed the usual “just bung it all in there with a couple of bit of styrofoam to fill the gaps” packing style, and sure enough some of the parts had dents and scratches.

The assembly was fairly trouble-free, although there was one screwhead that I cannot fathom how they expected someone with normal-sized hands to reach — maybe if I could have trusted Nicholas with my stubby Philips-head we might have got it tightened.

Then once I got it all together, Susan remarked “it’s not supposed to have white drawers, is it?”  Sure enough, they packed the wrong drawer fronts.  Contacting the company about the mistake, they said “you’ll have to wait until we can order the parts”.  When we offered to take the ones from display, something that shops will often do to stop a customer complaining, and they said “no, there will be people in the shop over the weekend looking at the desk” (meaning the customers who haven’t bought anything yet are more important than the ones that have?).

So to my thinking, this transaction has been well below the expectations I had set based on the price we paid for the item — but at each of the decision points during the transaction, the choice went in the shop’s direction.  Why?

This is the complex bit.  We paid $299 for the item, and factoring in the fact that we had to put it together ourselves I figure that in the traditional furniture model — where you buy finished furniture that is already assembled, and someone probably delivers for you as well — that would be something like $400-$450.  In my way of thinking, paying that amount of money for anything comes with an expectation that the level of service and quality would be considerably higher than your average supermarket transaction (more on that later).  Is the reason that we haven’t challenged the level of service influenced by a doubt that we really have paid enough to earn a higher level of service?  Maybe had we bought the $500 desk, or the $1000 one, I’d be commenting about the fantastic service instead of the crummy service…  I’m sure this is only going to get worse as I grow older and the effects of inflation start to really kick in.

Oh, the supermarket reference — Susan arrived home after a supermarket trip to find that some of her groceries were missing.  Checking the docket, she realised it was the last two items on the docket — a loaf of bread and a dozen eggs — didn’t make it home.  She contacted the supermarket (in spite of my heckling!) and to the surprise of both of us they said “yes, you did leave them behind, come and collect fresh ones when you next come to the store”.  It appears that they keep a book, imaginitively entitled “Stuff Customers Leave at the Checkouts”, into which the details of items left behind are recorded.  If someone contacts the shop to say they left something behind, they cross-reference it in the book and if there’s a matching entry the customer gets to pick up their forgotten items!

So sometimes the level of service can exceed the amount paid…  :)

A TOE involves me swallowing an ultrasound transducer (a tiny one designed for the job, not one like those used for foetal ultrasound, ladies!).  The transducer ends up in the stomach (which sits right beside the heart) allowing images to be taken of the heart without having to work past bone and muscle like a normal echocardiogram.

The specific condition I’m being tested for is patent foramen ovale (Wikipedia can tell you all about it here.  As far as atrial septal defects (the medical term for “hole-in-the-heart”) go it’s one of the more minor ones, but I guess if it’s worth testing for it’s worth treating.

The reason it might be relevant in my case is that it provides one possible explanation for the clots that might have caused my cardiac episodes.  Clots in the veins are quite common, but the veinous system carries them to the lungs and they generally don’t cause issues.  A hole-in-the-heart provides a path for clots to cross over into the arteries, where they can cause issues by causing blockage in coronary arteries (leading to heart attack), cerebral arteries (leading to stroke) or any other part of the body for that matter.

So I’m off to the hospital to have a tube rammed down my neck and get some more pictures taken of my heart.  I’m starting to feel a bit like a pincushion, but if we can get some answers it will be worthwhile.  Wish me luck!

In case you’re curious, the reason for the failure seems to be that emerge of your system and/or world trees doesn’t seem to handle slotted packages (in Gentoo, packages that can have more than one version installed at once, like the kernel source, GCC and others, are called slotted; each different version installs into a different ‘slot’ and can be managed differently).  GTK+ is slotted, and to satisfy dependencies for various packages I had both version 1 and 2 installed in different slots.  The emerge world only seemed to pick up one of them (version 1 by the look), so packages dependent on gtk+-2 were failing.

I’ve started doing the GCC/glibc update on a virtual server system I have.  Looks like it’s having similar problems, although I think it’s because I’m trying to switch to the new “server” profile from 2006.1 when I had already installed packages from the general profile.

I relocated directories for a couple of the internal webapps I run.  This was largely no problem, and mostly handled by telling webapp-config to upgrade the directory.  However, Cacti was a problem.  I had a mess of things to tidy up — the cron job that runs the poller, every single data source record (they had a hard-coded path to the RRD file).  Then, when it seemed like everything was fixed, I decided to fix up the display issue where Cacti picked up one of the alias IPs on my server Ethernet connection instead of the primary IP.  Suddenly, it could not see any data sources.

I tore out my hair checking all the database updates, config files, and so on — nothing.  So, I turned to Google, and in fairly short order came up with this link.  Looks like it’s always been this way, and was bound to cause some kind of problem eventually.  So I redefined the Cacti database tables like they said, et voila!

None of this had anything much to do with the SSL enablement question.  Well, when I was fighting with the network config on the server earlier, I added a couple of alias IP addresses.  Then in the Apache config I switched from using “VirtualHost *:443″ (which can’t work more than once, to which the error messages on startup attested) to specifying an IP for each domain.  Error messages gone, and everything works!

Now to get the reverse proxying config enabled so that I can present my newly-enabled SSL virtual hosts to the world!

This is a weird problem  My syslog-ng consoles just stopped writing, although things like ircd were still running.  Full /var?  The first thing I though of, but it was only 80% used (unless that’s after a logrotate cleanup, and /var had actually filled during the night).

Things got especially weird when I tried to reboot.  My eth0 interface did not start properly, and eth1, which is not properly configured, was starting instead.  Then, various other services failed to start, which seemed to hang init — and with init hanging, my console gettys were not being started.  Thank heaven that somewhere along the line the config of KDM was changed to start it on the local console, allowing me to log on that way…

I had enabled bootchart, and I thought it was the problem.  There’s definitely a problem with bootchart, as all the error messages it spews out when it fills the tempfs it creates cannot be normal, but it wasn’t what was stopping my boot.

I had also enabled the parallel startup feature of Gentoo’s init, but that was not the issue either (although it did give me a somewhat clearer picture of the services that were not starting properly).

The eventual problem with the network startup was the introduction of coldplug support to the Gentoo boot process.  Both the Gigabit interfaces on the server are managed by the one driver, and udev must process events by LIFO (last-in-first-out) so it saw eth1 before eth0.  Then the kicker — after it thought it had brought up eth1, it stopped trying to bring up more interfaces because the config only required one interface other than lo to be up.

The fix, as documented out there on the Innernut, is to turn off coldplug and/or restrict it from handling the services representing the network interfaces.

So now all I need to do is undo all the undoing I did to try and sort the bootup problem (re-enable parallel startup, reinstall bootchart and fix the bug).  Oh, that’s right, the version of baselayout that introduced all these nice changes also undid the dependency fix I made that started OpenLDAP nice ane early in the boot process, so now it takes ages to boot again (I never get tired of seeing the LDAP server not start properly because the LDAP server is not available…).