From matt at ucc.asn.au Sun Apr 1 17:32:03 2007 From: matt at ucc.asn.au (Matt Johnston) Date: Sun, 1 Apr 2007 17:32:03 +0800 Subject: [tech] Webserving moved to mussel Message-ID: <20070401093202.GK5332@ucc.gu.uwa.edu.au> Hi all. Since mermaid is slow and sucky, I've migrated all of UCC's webserving to mussel. It's apache2 with basically the same setup - everything I've checked works, but I guess some user webpages might have issues. The biggest change is probably that it's running php5 rather than 4. If anyone sees something awry, please let me know. I'll keep an eye on error.log too. Moving to mussel also is a step closer to pointing www.ucc (and ucc. as well I guess) directly at the webserver so cookies/virtualhosts etc work a lot more nicely. That can wait though I guess - anyone see problems with that plan? Matt From davyd at madeley.id.au Sun Apr 1 17:44:47 2007 From: davyd at madeley.id.au (Davyd Madeley) Date: Sun, 01 Apr 2007 17:44:47 +0800 Subject: [tech] Webserving moved to mussel In-Reply-To: <20070401093202.GK5332@ucc.gu.uwa.edu.au> References: <20070401093202.GK5332@ucc.gu.uwa.edu.au> Message-ID: <1175420687.16203.0.camel@frobisher.madeley.id.au> On Sun, 2007-04-01 at 17:32 +0800, Matt Johnston wrote: > Moving to mussel also is a step closer to pointing www.ucc > (and ucc. as well I guess) directly at the webserver so > cookies/virtualhosts etc work a lot more nicely. That can > wait though I guess - anyone see problems with that plan? I didn't think we reverse-proxied for speed. --d -- Davyd Madeley http://www.davyd.id.au/ 08B0 341A 0B9B 08BB 2118 C060 2EDD BB4F 5191 6CDA From matt at ucc.asn.au Sun Apr 1 17:53:31 2007 From: matt at ucc.asn.au (Matt Johnston) Date: Sun, 1 Apr 2007 17:53:31 +0800 Subject: [tech] Webserving moved to mussel In-Reply-To: <1175420687.16203.0.camel@frobisher.madeley.id.au> References: <20070401093202.GK5332@ucc.gu.uwa.edu.au> <1175420687.16203.0.camel@frobisher.madeley.id.au> Message-ID: <20070401095331.GL5332@ucc.gu.uwa.edu.au> On Sun, Apr 01, 2007 at 05:44:47PM +0800, Davyd Madeley wrote: > On Sun, 2007-04-01 at 17:32 +0800, Matt Johnston wrote: > > > Moving to mussel also is a step closer to pointing www.ucc > > (and ucc. as well I guess) directly at the webserver so > > cookies/virtualhosts etc work a lot more nicely. That can > > wait though I guess - anyone see problems with that plan? > > I didn't think we reverse-proxied for speed. Nah. But given there's some level of hassle involved in changing the IP of the server, we may as well move it to one that's actually non-sucky first. I'd envision that we'd use .13.28 for the main webserver (currently secure.ucc as an extra interface on mussel), since that's already set up sanely for allowing connections in but not establishing outbound. I guess we'd add a bright IP too so that the bulk-www uncharged traffic trick works. A note for those who haven't seen - the host bulk-www.ucc.asn.au works the same as www.ucc.asn.au, but we don't get charged per-meg for backtraffic. So if you have large images or similar in your webspace, that URL is more club-friendly. Something such as this in your ~/public-html/.htaccess will make files redirect automagically: (change the username of course) RewriteBase /~matt # some downloaders such as curl don't like 302s - keep a whitelist RewriteCond %{HTTP_USER_AGENT} ^(Mozilla|msnbot|Wget|Opera).* RewriteCond %{HTTP:X-Forwarded-Host} !(bulk-www.*|ext-mx.*) RewriteRule ^(minitar|mirror|dropbear|xxdiff|monotone|nowire|files|memtest)/(. *\.(jpg|jpeg|gif|png|gz|bz2|img|pdf|ps|exe|zip|dmg|swf|deb|avi|iso)$) http://bu lk-www.ucc.asn.au/~matt/$1/$2 [R] Matt From shmookey at shmookey.net Sun Apr 1 18:03:12 2007 From: shmookey at shmookey.net (Luke Williams) Date: Sun, 1 Apr 2007 18:03:12 +0800 Subject: [tech] Webserving moved to mussel In-Reply-To: <20070401093202.GK5332@ucc.gu.uwa.edu.au> References: <20070401093202.GK5332@ucc.gu.uwa.edu.au> Message-ID: On 4/1/07, Matt Johnston wrote: > If anyone sees something awry, please let me know. I'll keep > an eye on error.log too. Python CGI files aren't working anymore for me on my webspace. They were working earlier today. luke From davyd at madeley.id.au Mon Apr 2 09:33:36 2007 From: davyd at madeley.id.au (Davyd Madeley) Date: Mon, 02 Apr 2007 09:33:36 +0800 Subject: [tech] Webserving moved to mussel In-Reply-To: <20070401093202.GK5332@ucc.gu.uwa.edu.au> References: <20070401093202.GK5332@ucc.gu.uwa.edu.au> Message-ID: <1175477618.7152.1.camel@frobisher.madeley.id.au> On Sun, 2007-04-01 at 17:32 +0800, Matt Johnston wrote: > If anyone sees something awry, please let me know. I'll keep > an eye on error.log too. The logs on mussel are owned by adm, not www-data (a little bit annoying). Also, DirectoryIndex appear to be set wrong, preferring the index.html to the index.ucc. --d -- Davyd Madeley http://www.davyd.id.au/ 08B0 341A 0B9B 08BB 2118 C060 2EDD BB4F 5191 6CDA From trs80 at ucc.gu.uwa.edu.au Tue Apr 3 17:03:23 2007 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Tue, 3 Apr 2007 17:03:23 +0800 (WST) Subject: [tech] Webserving moved to mussel In-Reply-To: <20070401093202.GK5332@ucc.gu.uwa.edu.au> References: <20070401093202.GK5332@ucc.gu.uwa.edu.au> Message-ID: On Sun, 1 Apr 2007, Matt Johnston wrote: > If anyone sees something awry, please let me know. I'll keep > an eye on error.log too. I changed the ProxyPass on mooneye to point to realwww, which is a CNAME to mussel, and added a separate VirtualHost section for mussel.ucc with a DocumentRoot of /var/www, so the stuff there works again. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From trs80 at ucc.gu.uwa.edu.au Mon Apr 23 23:31:31 2007 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Mon, 23 Apr 2007 23:31:31 +0800 (WST) Subject: [tech] upgrades Message-ID: I upgraded madako to etch tonight, and manbo's blastwave packages today. Someone should probably check madako's apache2 setup. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From trs80 at ucc.gu.uwa.edu.au Fri Apr 27 08:32:10 2007 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Fri, 27 Apr 2007 08:32:10 +0800 (WST) Subject: [tech] mooneye, mermaid upgrades In-Reply-To: References: Message-ID: On Fri, 27 Apr 2007, James Andrewartha wrote: > On Fri, 27 Apr 2007, James Andrewartha wrote: > >> On Thu, 26 Apr 2007, James Andrewartha wrote: >> >>> So I upgraded mooneye and mermaid tonight. Everything seemed to go >>> smoothly, apart from me forgetting to update the grub config on mermaid to >>> use udev paths instead of devfs ones. Things that probably need to be >>> looked at in more detail: dovecot on mooneye, merging in upstream's >>> apache2.conf on mooneye, mermaid too if anyone cares. >> >> Well, except that I broke mail overnight - a combination of libnss-ldap in >> etch sucking and the localnorelay hack needing updating. libnss-ldap is >> downgraded, but localnorelay will need fixing. > > And mailman needed s/python2.3/python2.4/ Mmmm, editing binaries. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From trs80 at ucc.gu.uwa.edu.au Fri Apr 27 15:54:37 2007 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Fri, 27 Apr 2007 15:54:37 +0800 (WST) Subject: [tech] mooneye, mermaid upgrades In-Reply-To: References: Message-ID: On Fri, 27 Apr 2007, James Andrewartha wrote: >>>> So I upgraded mooneye and mermaid tonight. Everything seemed to go >>>> smoothly, apart from me forgetting to update the grub config on mermaid to >>>> use udev paths instead of devfs ones. Things that probably need to be >>>> looked at in more detail: dovecot on mooneye, merging in upstream's >>>> apache2.conf on mooneye, mermaid too if anyone cares. >>> >>> Well, except that I broke mail overnight - a combination of libnss-ldap in >>> etch sucking and the localnorelay hack needing updating. libnss-ldap is >>> downgraded, but localnorelay will need fixing. >> >> And mailman needed s/python2.3/python2.4/ > > Mmmm, editing binaries. More binary editing in mailman/cgi-bin. Serial was broken on mermaid too, but that's now fixed, however coke still isn't working and I can't talk to the machine via minicom, might be a layer 1 issue. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From trs80 at ucc.gu.uwa.edu.au Fri Apr 27 15:56:14 2007 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Fri, 27 Apr 2007 15:56:14 +0800 (WST) Subject: [tech] mooneye, mermaid upgrades In-Reply-To: References: Message-ID: On Fri, 27 Apr 2007, James Andrewartha wrote: > More binary editing in mailman/cgi-bin. Serial was broken on mermaid too, > but that's now fixed, however coke still isn't working and I can't talk to > the machine via minicom, might be a layer 1 issue. The other thing I meant to mention was that the UCC theme for the wiki needs updating, things like the edit page are bare. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From dagobah at ucc.gu.uwa.edu.au Fri Apr 27 22:02:03 2007 From: dagobah at ucc.gu.uwa.edu.au (Bernard Blackham) Date: Fri, 27 Apr 2007 22:02:03 +0800 Subject: [tech] mooneye, mermaid upgrades In-Reply-To: References: Message-ID: <4632025B.6040605@ucc.gu.uwa.edu.au> > Serial was broken on mermaid too, > but that's now fixed, however coke still isn't working and I can't talk to > the machine via minicom, might be a layer 1 issue. As per http://wiki.ucc.asn.au/FixCoke , it now works again. Bernard. From adrian at ucc.gu.uwa.edu.au Sat Apr 28 11:00:36 2007 From: adrian at ucc.gu.uwa.edu.au (Adrian Chadd) Date: Sat, 28 Apr 2007 11:00:36 +0800 Subject: [tech] manbo gige? Message-ID: <20070428030036.GF24585@ucc.gu.uwa.edu.au> Hiya, Can anyone remember what optics/whatnot Manbo needs to use gigabit ethernet? I've got GBIC/SFP modules available.. Adrian From zanchey at ucc.gu.uwa.edu.au Sat Apr 28 17:12:48 2007 From: zanchey at ucc.gu.uwa.edu.au (David Adam) Date: Sat, 28 Apr 2007 17:12:48 +0800 (WST) Subject: [tech] mooneye, mermaid upgrades In-Reply-To: <4632025B.6040605@ucc.gu.uwa.edu.au> References: <4632025B.6040605@ucc.gu.uwa.edu.au> Message-ID: On Fri, 27 Apr 2007, Bernard Blackham wrote: > > Serial was broken on mermaid too, > > but that's now fixed, however coke still isn't working and I can't talk to > > the machine via minicom, might be a layer 1 issue. > > As per http://wiki.ucc.asn.au/FixCoke , it now works again. Interestingly, I couldn't get 'rmmod 8250 && modprobe 8250' to work the other day. The module wasn't loaded, and then modprobe failed with 'device in use' or something similar. David Adam UCC Wheel Member zanchey at ucc.gu.uwa.edu.au From adrian at ucc.gu.uwa.edu.au Sun Apr 29 17:13:32 2007 From: adrian at ucc.gu.uwa.edu.au (Adrian Chadd) Date: Sun, 29 Apr 2007 17:13:32 +0800 Subject: [tech] manbo gige Message-ID: <20070429091332.GA14325@ucc.gu.uwa.edu.au> Trs and I made manbo gigabit-happy. I threw in a cisco SX Gbic into the 2948G (which I've christened "vegetable" to go with "olive".) The only way to get the thing going was to disable link negotiation on both ends. TRS has to add that into /etc/system or whatnot or manbo's gigabit interface won't come back up on boot. The interface name is ge1 and its on board gem0. This caused no end of confusion as we were trying to bring the thing up. The sun card refused to bring up carrier on the port until we plumbed the interface. Apparently this board performs rather horribly as it requires a bit of CPU intervention to push bits around. I believe the actual cause of this is the lack of offload processing (TSO/checksum, etc.) I wonder if this'll manifest itself on manbo. Yes, UCC can keep the GBIC. I have an SX SFP module for the alloy switch and/or whatever replacement we buy. Someone just needs to acquire a LC to SC Multimode cable or two. UCC can have that too if they replace the alloy. :) Enjoy. Adrian From adrian at ucc.gu.uwa.edu.au Sun Apr 29 17:15:14 2007 From: adrian at ucc.gu.uwa.edu.au (Adrian Chadd) Date: Sun, 29 Apr 2007 17:15:14 +0800 Subject: [tech] cacti Message-ID: <20070429091514.GB14325@ucc.gu.uwa.edu.au> I've installed cacti on mussel - http://mussel.ucc.asn.au/cacti/ . Cacti is horribly insecure and as such I'll hide it behind htpasswd access soon. The admin login is the same as the switch enable password. I forget whether I capitalised the F or not. I'll hide cacti behind the htpasswd with user admin and the switch password. I'm graphing 64 bit bit counters and errors/discards on both olive and vegetable. Adrian From adrian at ucc.gu.uwa.edu.au Sun Apr 29 17:32:11 2007 From: adrian at ucc.gu.uwa.edu.au (Adrian Chadd) Date: Sun, 29 Apr 2007 17:32:11 +0800 Subject: [tech] cacti In-Reply-To: <20070429091514.GB14325@ucc.gu.uwa.edu.au> References: <20070429091514.GB14325@ucc.gu.uwa.edu.au> Message-ID: <20070429093210.GC14325@ucc.gu.uwa.edu.au> On Sun, Apr 29, 2007, Adrian Chadd wrote: > > I've installed cacti on mussel - http://mussel.ucc.asn.au/cacti/ . > Cacti is horribly insecure and as such I'll hide it behind htpasswd > access soon. > > The admin login is the same as the switch enable password. > I forget whether I capitalised the F or not. I'll hide cacti > behind the htpasswd with user admin and the switch password. > > I'm graphing 64 bit bit counters and errors/discards on both > olive and vegetable. olive doesn't do 64 bit counters it seems; so olive's port graphs are now 32 bit. Adrian From trs80 at ucc.gu.uwa.edu.au Sun Apr 29 18:40:06 2007 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Sun, 29 Apr 2007 18:40:06 +0800 (WST) Subject: [tech] mooneye, mermaid upgrades In-Reply-To: References: <4632025B.6040605@ucc.gu.uwa.edu.au> Message-ID: On Sat, 28 Apr 2007, David Adam wrote: >>> Serial was broken on mermaid too, but that's now fixed, however coke >>> still isn't working and I can't talk to the machine via minicom, might >>> be a layer 1 issue. >> >> As per http://wiki.ucc.asn.au/FixCoke , it now works again. > > Interestingly, I couldn't get 'rmmod 8250 && modprobe 8250' to work the > other day. The module wasn't loaded, and then modprobe failed with > 'device in use' or something similar. Yeah, I had to reconfigure xen's console to not grab the serial posts - set console=vga in the kernel line, and xencons=tty in the module line. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From trs80 at ucc.gu.uwa.edu.au Sun Apr 29 18:54:01 2007 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Sun, 29 Apr 2007 18:54:01 +0800 (WST) Subject: [tech] manbo gige In-Reply-To: <20070429091332.GA14325@ucc.gu.uwa.edu.au> References: <20070429091332.GA14325@ucc.gu.uwa.edu.au> Message-ID: On Sun, 29 Apr 2007, Adrian Chadd wrote: > The only way to get the thing going was to disable link negotiation on both > ends. TRS has to add that into /etc/system or whatnot or manbo's gigabit > interface won't come back up on boot. Actually it does autoneg fine, so no boot-time hacks needed. > The interface name is ge1 and its on board gem0. This caused no end of > confusion as we were trying to bring the thing up. The sun card refused > to bring up carrier on the port until we plumbed the interface. It is gem1, but why it's not ge0/gem0 I have no idea. -- # TRS-80 trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \ # UCC Wheel Member http://trs80.ucc.asn.au/ #| what squirrels do best | [ "There's nobody getting rich writing ]| -- Collect and hide your | [ software that I know of" -- Bill Gates, 1980 ]\ nuts." -- Acid Reflux #231 / From splintax at ucc.asn.au Mon Apr 30 19:39:19 2007 From: splintax at ucc.asn.au (Scott Young) Date: Mon, 30 Apr 2007 19:39:19 +0800 Subject: [tech] Problems with arctic (eMac) Message-ID: <2bc799480704300439i20d0390bga0ccbc7d5e007a8f@mail.gmail.com> Hi all, I've noticed that even after Luke wiped and reinstalled OS X, the Mac still seems to be performing very poorly. I plugged in my iPod to do some work today (had documents saved on it) and upon automatically loading, iTunes crashed with an out-of-memory error. Programs like MS Word are painfully slow to load and even stuff like Terminal.app takes several bounces to come up. :-( I'd like to at least look into getting some basic diagnostic tests done on it but not being much of a Mac person myself I'm not sure where to start. (Also, I'm not on wheel.) This seems to be the go as far as a version of memtest86 that we can run on PPC Macs: http://www.memtestosx.org Does anybody else have any suggestions on what could possibly be the problem with the machine? I'm aware that it's "getting old", but out of memory errors shouldn't be happening, especially seeing as it has a gig of RAM.. -Scott [splintax] From harrymc at ucc.gu.uwa.edu.au Mon Apr 30 21:05:57 2007 From: harrymc at ucc.gu.uwa.edu.au (Harry) Date: Mon, 30 Apr 2007 21:05:57 +0800 Subject: [tech] AM100 progress Message-ID: <4635E9B5.5010804@ucc.gu.uwa.edu.au> Hello This evening I moved the ex-WMC, Bernd-donated AM100 up onto the hardware bench after returning tools to the cupboard and a general tidy up. The main 50000uF caps had been running for several days wired independently to a 5 volt supply: 150K 5V---/\/\/\------| | = 50000uF | GND--------------| Terminal capacitor voltages were 4.95V and 4.40V respectively with a drop across the resistors of 71.2mV and 627mV representing leakage currents of 0.5 microamps and 4.2 microamps. Despite the difference in leakage, I think the caps are ok. The lower leakage capacitor showed no voltage loss after several minutes when loaded with a (10Mohm) multimeter whereas the leakier capacitor lost about 4 millivolts per minute; which is self discharge from the internal leakage. These capacitors still hold a significant charge so please don't tinker with the wiring in the box. I haven't removed the S100 boards yet so wire twiddling may make fat sparks or partially power the boards both of which would be bad. I've set up two 1 Mohm resistors tonight to independently charge the smaller capacitors with a +/- 12 volts source. These supplies provide the unregulated +/- 15 volt supplies (regulated on the S100 boards to +/-12 volts mainly for RS232 drivers). I'll be back tomorrow in the early afternoon to check on those. These gradual charges are at very small currents so the likelyhood of catastrophic failure is low. After this charging, I'll restore the wiring to the capacitors, remove the S100 boards and use some wirewound resistors as a dummy load and see how the power supplies stand up. So power to the system to test for life is a few steps away yet. Harry From shmookey at shmookey.net Mon Apr 30 21:09:54 2007 From: shmookey at shmookey.net (Luke Williams) Date: Mon, 30 Apr 2007 21:09:54 +0800 Subject: [tech] Problems with arctic (eMac) In-Reply-To: <2bc799480704300439i20d0390bga0ccbc7d5e007a8f@mail.gmail.com> References: <2bc799480704300439i20d0390bga0ccbc7d5e007a8f@mail.gmail.com> Message-ID: On 4/30/07, Scott Young wrote: > I've noticed that even after Luke wiped and reinstalled OS X, the Mac > still seems to be performing very poorly. I did no such thing! From splintax at ucc.asn.au Mon Apr 30 23:09:31 2007 From: splintax at ucc.asn.au (Scott Young) Date: Mon, 30 Apr 2007 23:09:31 +0800 Subject: [tech] Problems with arctic (eMac) In-Reply-To: References: <2bc799480704300439i20d0390bga0ccbc7d5e007a8f@mail.gmail.com> Message-ID: <2bc799480704300809w156f2788kd0ac545ae84093cc@mail.gmail.com> I meant to check that with you before posting that. My mistake. I remember you were planning on doing it, and even had the discs and everything ready, correct? There was a problem with getting the machine to boot off the CD, right? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.ucc.gu.uwa.edu.au/pipermail/tech/attachments/20070430/3c601548/attachment.htm From matt at ucc.asn.au Mon Apr 30 23:37:55 2007 From: matt at ucc.asn.au (Matt Johnston) Date: Mon, 30 Apr 2007 23:37:55 +0800 Subject: [tech] Problems with arctic (eMac) In-Reply-To: <2bc799480704300809w156f2788kd0ac545ae84093cc@mail.gmail.com> References: <2bc799480704300439i20d0390bga0ccbc7d5e007a8f@mail.gmail.com> <2bc799480704300809w156f2788kd0ac545ae84093cc@mail.gmail.com> Message-ID: <20070430153755.GC27607@ucc.gu.uwa.edu.au> On Mon, Apr 30, 2007 at 11:09:31PM +0800, Scott Young wrote: > I meant to check that with you before posting that. My mistake. I remember > you were planning on doing it, and even had the discs and everything ready, > correct? > > There was a problem with getting the machine to boot off the CD, right? Nah, the reason was that Alex and I had already reinstalled it at the Black Friday Lan a few weeks back. My conjecture is that sucky NFS performance is the root cause of much slowness. Not sure if we can tweak/recable to improve the NFS, or whether maybe we should try Samba or even netatalkd (or maybe not). I'm doubting it's bad memory/disk, but it couldn't hurt running memtestosx (though it's pay-for-gpl) and also the hardware test CD that came with it. Not sure about iTunes, I couldn't see anything in the logs. Though Word was complaining about some broken resource fork type thing - it probably should be reinstalled? It'll be quicker starting up first time since it won't install fonts for each user (maybe saving space too?). Matt