From trs80 at ucc.gu.uwa.edu.au Tue Jun 10 16:04:16 2008 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Tue, 10 Jun 2008 16:04:16 +0800 (WST) Subject: [tech] aircon Message-ID: As you may have noticed from the temperature graphs, the aircon circuit tripped again, at about 9:20pm last night. This is the second time it's done this, it also went at about 6am last Friday. -- # 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 harrymc at ucc.gu.uwa.edu.au Wed Jun 11 09:22:36 2008 From: harrymc at ucc.gu.uwa.edu.au (Harry) Date: Wed, 11 Jun 2008 09:22:36 +0800 Subject: [tech] aircon In-Reply-To: References: Message-ID: <484F28DC.6090800@ucc.gu.uwa.edu.au> James Andrewartha wrote: > As you may have noticed from the temperature graphs, the aircon circuit > tripped again, at about 9:20pm last night. This is the second time it's > done this, it also went at about 6am last Friday. Dear UCC Could UCC get the brand/model of the aircon Zanchey says has been 24/7 reliable at his work in a machine room and investigate fabricating a bracket to hang the two parts each side of the window ledge ? No tinkering with the heritage listed CH walls and the windows can be closed (except the last two inches) to let light into the MR. I know this sounds very hand wavy but I think it'd work quite well structurally and make the aircon issues just go away. Love, Bob the Builder ps I'm up for building a shiny new MR as well but that's too much fixing for one episode. From frenchie at ucc.gu.uwa.edu.au Fri Jun 13 13:17:27 2008 From: frenchie at ucc.gu.uwa.edu.au (James French) Date: Fri, 13 Jun 2008 13:17:27 +0800 Subject: [tech] [wheel] Fwd: [Tech-contacts] Apple Software Update Server References: <980B4CBA-3A9D-4F11-ACC2-E50B097301EE@uwa.edu.au> Message-ID: <8E119E36-0300-45ED-B2B8-0A5FDB1AF853@ucc.gu.uwa.edu.au> I have set this up on Apollo and Arctic. Although we rarely get the traffic from non-free sites, just guarantees that the updates will always come down at a sensible rate. Fwiw I also set up the software update server and know it's not doing anything crazy... well, anything crazy for OS X server, which is to say, lots of things crazy. Frenchie Begin forwarded message: > From: Alex Dawson > Date: 13 June 2008 11:33:30 AM > To: tech-contacts at uwa.edu.au > Subject: [Tech-contacts] Apple Software Update Server > > Hi everyone, > > I was asked by someone about an on-campus Apple Software Update > Server. While in most cases, the traffic to the official Apple ones > will be free, due to Akamai/iinet/WAIX, in some cases it may not, > and firewall rules may block direct access to the Internet in some > areas. > > In any case, we have got one at http://asu.arts.uwa.edu.au:8088 .. > holding updates for 10.4 and 10.5, and anyone on campus is welcome > to use it, on the proviso that you let us know. To set it up, we > have the following script: > > --- 8< --- > > #!/bin/sh > > SUSERVER="http://asu.arts.uwa.edu.au:8088/" # trailing backslash is > REQUIRED! > DWRITE="/usr/bin/defaults write" > > # Point at local Software Update server > $DWRITE "/Library/Preferences/com.apple.SoftwareUpdate" CatalogURL "$ > {SUSERVER}" > > --- 8< --- > > Regards, > -- > Alex Dawson > FAHSS Faculty Computing, University of Western Australia (CRICOS > 00126G) > 08 6488 7093 - alex.dawson at uwa.edu.au - http://uwa.edu.au/people/alex.dawson > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.ucc.gu.uwa.edu.au/pipermail/tech/attachments/20080613/1f9e39ec/attachment.htm From zanchey at ucc.gu.uwa.edu.au Fri Jun 13 15:43:38 2008 From: zanchey at ucc.gu.uwa.edu.au (David Adam) Date: Fri, 13 Jun 2008 15:43:38 +0800 (WST) Subject: [tech] [wheel] Fwd: [Tech-contacts] Apple Software Update Server In-Reply-To: <8E119E36-0300-45ED-B2B8-0A5FDB1AF853@ucc.gu.uwa.edu.au> References: <980B4CBA-3A9D-4F11-ACC2-E50B097301EE@uwa.edu.au> <8E119E36-0300-45ED-B2B8-0A5FDB1AF853@ucc.gu.uwa.edu.au> Message-ID: On Fri, 13 Jun 2008, James French wrote: > I have set this up on Apollo and Arctic. Although we rarely get the traffic > from non-free sites, just guarantees that the updates will always come down at > a sensible rate. > > Fwiw I also set up the software update server and know it's not doing anything > crazy... well, anything crazy for OS X server, which is to say, lots of things > crazy. Funnily enough I was just thinking about Apple system updates today. We have automatic updates enabled on all the Windows machines (or we should have), phonehome for the Debian boxes, pkg-update-watch for the CSW stuff on Solaris... softwareupdate(8) on Mac OS X is a bit underpowered, but something like the script below might be useful. Alternatively, have any of the Mac-heads on the list got a better suggestion for keeping track of software updates on our two Macs? We could just cron 'softwareupdate -i --all && reboot' but that seems a bit dangerous. I guess we could always port Phonehome... David Adam UCC Wheel Member zanchey at ucc.gu.uwa.edu.au --- #! /bin/sh OUTPUT=`softwareupdate -l` if (echo $OUTPUT | grep "found the following new or updated software"); then mail -s "Software updates available on $HOST" root at ucc < References: <980B4CBA-3A9D-4F11-ACC2-E50B097301EE@uwa.edu.au> <8E119E36-0300-45ED-B2B8-0A5FDB1AF853@ucc.gu.uwa.edu.au> Message-ID: 2008/6/13 David Adam : > I guess we could always port Phonehome... Phonehome is just a few Python scripts and deliberately simple, writing a plugin should be pretty trivial. Anyone that wants to do it, feel free to commit to UCC's monotone server on the branch :-) From frenchie at ucc.gu.uwa.edu.au Sun Jun 15 11:26:31 2008 From: frenchie at ucc.gu.uwa.edu.au (James French) Date: Sun, 15 Jun 2008 11:26:31 +0800 Subject: [tech] Breaking out some of Mussel's services. Message-ID: <560b010806142026p44d49d45hffa960dea97da244@mail.gmail.com> Do we want to move some of the core services mussel has running onto another machine/VM. I'm all for keeping Mussel a general user machine running general user services but when random service B takes the machine down it's annoying. Specifically, I'm thinking LDAP, IMAP/POP and Web could stand to be put on a different machine. LDAP is the biggest problem as when it goes away it makes getting in remotely a royal pita. Lastly, before I get accused of the opposite I do like the fact that we don't just run web, mail and DNS. Mussel is a great place to run the non-core stuff like games servers. What do other people think? Frenchie -- James French frenchie at ucc.gu.uwa.edu.au From grahame at angrygoats.net Mon Jun 16 00:11:16 2008 From: grahame at angrygoats.net (Grahame Bowland) Date: Mon, 16 Jun 2008 00:11:16 +0800 Subject: [tech] Breaking out some of Mussel's services. In-Reply-To: <560b010806142026p44d49d45hffa960dea97da244@mail.gmail.com> References: <560b010806142026p44d49d45hffa960dea97da244@mail.gmail.com> Message-ID: Hey Does the slapd replication stuff work these days? It used to be a real pain in the arse to get working, I think I managed to get it going once. If we had a backup LDAP server, we'd be able to get in (although I guess you'd have to wait for query to the dead LDAP primary server to time out.) Grahame 2008/6/15 James French : > Do we want to move some of the core services mussel has running onto > another machine/VM. I'm all for keeping Mussel a general user machine > running general user services but when random service B takes the > machine down it's annoying. > > Specifically, I'm thinking LDAP, IMAP/POP and Web could stand to be > put on a different machine. LDAP is the biggest problem as when it > goes away it makes getting in remotely a royal pita. > > Lastly, before I get accused of the opposite I do like the fact that > we don't just run web, mail and DNS. Mussel is a great place to run > the non-core stuff like games servers. > > What do other people think? > > Frenchie > > -- > James French > frenchie at ucc.gu.uwa.edu.au > > > From frenchie at ucc.gu.uwa.edu.au Mon Jun 16 00:16:36 2008 From: frenchie at ucc.gu.uwa.edu.au (James French) Date: Mon, 16 Jun 2008 00:16:36 +0800 Subject: [tech] Breaking out some of Mussel's services. In-Reply-To: References: <560b010806142026p44d49d45hffa960dea97da244@mail.gmail.com> Message-ID: <560b010806150916u3b5f4b8bwf19c595c08e9328@mail.gmail.com> > Does the slapd replication stuff work these days? It used to be a real > pain in the arse to get working, I think I managed to get it going > once. If we had a backup LDAP server, we'd be able to get in (although > I guess you'd have to wait for query to the dead LDAP primary server > to time out.) We have martello doing backup, not sure what the status is, someone did mention something about SSL biting us. Trs80, Zanchey comments? It's not exactly hard to break those services out onto another VM. Last I checked we still had some resources available on Mermaid. Even if we have to scale back Mussel's resources slightly and gave those to another machine, that'd get us past the out of memory issues (though not away from the CPU getting thrashed). Frenchie -- James French frenchie at ucc.gu.uwa.edu.au From trs80 at ucc.gu.uwa.edu.au Mon Jun 16 01:39:12 2008 From: trs80 at ucc.gu.uwa.edu.au (James Andrewartha) Date: Mon, 16 Jun 2008 01:39:12 +0800 (WST) Subject: [tech] Breaking out some of Mussel's services. In-Reply-To: <560b010806150916u3b5f4b8bwf19c595c08e9328@mail.gmail.com> References: <560b010806142026p44d49d45hffa960dea97da244@mail.gmail.com> <560b010806150916u3b5f4b8bwf19c595c08e9328@mail.gmail.com> Message-ID: On Mon, 16 Jun 2008, James French wrote: > > Does the slapd replication stuff work these days? It used to be a real > > pain in the arse to get working, I think I managed to get it going > > once. If we had a backup LDAP server, we'd be able to get in (although > > I guess you'd have to wait for query to the dead LDAP primary server > > to time out.) > > We have martello doing backup, not sure what the status is, someone > did mention something about SSL biting us. Trs80, Zanchey comments? We have slapd replication, the issue is PAM failover doesn't work when doing LDAP with TLS. This may be fixed in lenny. > It's not exactly hard to break those services out onto another VM. > Last I checked we still had some resources available on Mermaid. Even > if we have to scale back Mussel's resources slightly and gave those to > another machine, that'd get us past the out of memory issues (though > not away from the CPU getting thrashed). I've moved the steam server onto martello, it was the most resource-hungry thing on mussel (slapd itself coming a close second). IMAP/POP are on mooneye, they were affected by the broken LDAP failover. -- # 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 matt at ucc.asn.au Mon Jun 16 23:52:37 2008 From: matt at ucc.asn.au (Matt Johnston) Date: Mon, 16 Jun 2008 23:52:37 +0800 Subject: [tech] Breaking out some of Mussel's services. In-Reply-To: <560b010806142026p44d49d45hffa960dea97da244@mail.gmail.com> References: <560b010806142026p44d49d45hffa960dea97da244@mail.gmail.com> Message-ID: <20080616155237.GF29723@ucc.gu.uwa.edu.au> On Sun, Jun 15, 2008 at 11:26:31AM +0800, James French wrote: > Do we want to move some of the core services mussel has running onto > another machine/VM. I'm all for keeping Mussel a general user machine > running general user services but when random service B takes the > machine down it's annoying. > > Specifically, I'm thinking LDAP, IMAP/POP and Web could stand to be > put on a different machine. LDAP is the biggest problem as when it > goes away it makes getting in remotely a royal pita. > > Lastly, before I get accused of the opposite I do like the fact that > we don't just run web, mail and DNS. Mussel is a great place to run > the non-core stuff like games servers. > > What do other people think? Nah. Just use ulimit -d on resource hog services already. And turn off SSL for ldap on core machines, then failover should work (hooray it trying to keep using the same SSL session to a different host). Matt