[12:38] <lamont__> jbailey: what determines what modules are loaded into the initrd?
[01:07] <infinity> lamont : Did you clean up ross?
[01:08] <infinity> lamont : I got a message from elmo last night that it was out of disk space; seems (more than) fine now..
[01:09] <elmo> [01-08-2005 09:46:18]  SERVICE ALERT: ross;Disk Space of /;CRITICAL;HARD;3;DISK CRITICAL [39041 kB (3%) free on /dev/sda3] 
[01:09] <elmo> I wasn't making it up :-P
[01:10] <infinity> I assume not. :)
[01:10] <infinity> Also, speaking of ross, it seems confused about SSH keys... Hrm.
[01:11] <elmo> yeah, it will be, it's hostname is fucked
[01:11] <infinity> I have 4 keys in LDAP (all show up in /var/lib/misc just fine, so ud-ldap is syncing okay), but I can only auth from one host, not the other.
[01:11] <elmo> try now
[01:11] <infinity> Oo.  Much better.
[01:11] <infinity> Was it authenticating against the hdb.com stuff?
[01:12] <elmo> actually ross has done this for like the last 5 or 6 days in a row
[01:12] <elmo> i.e. dropped to 2-5% free at ~9am local
[01:12] <elmo> so whatever's happening then is Bad(tm)
[01:13] <infinity> Evil cronjob?
[01:13] <elmo> infinity: yah, if you find a buildd not called .buildd (in hostname -f), yell
[01:13] <infinity> Yeah, I think there are a few others that hate me.  I'll loop through all 12 and see.
[01:17] <infinity> crested and floe seem to be the only other two.
[01:18] <infinity> 9am local, you say?  Hrm.
[01:18] <elmo> fixed
[01:19] <infinity> The only possibly nasty cronjob I can see if the DI build (which shouldn't be nasty anyway) at 6:15...
[01:19] <infinity> I guess if it's going out of control, it could take 3 hours to fill the disks.
[01:19] <infinity> s/if the/is the/
[01:20] <elmo> I'm fairly sure the d-i daily build is finished well before then
[01:20] <elmo> is ross not live CD image builder?
[01:20] <infinity> Nope.
[01:20] <elmo> damn, that was my first thought
[01:20] <infinity> The LiveCD builds are commented out on ross.
[01:21] <elmo> acct's useless for this sort of thing - unless lamont knows, I guess we should cron some ps auxfw's
[01:53] <lamont> ross is d-i
[01:54] <lamont> hrm.. clearly someone cleaned it up
[01:54] <lamont> royal is livecd
[01:54] <infinity> Not I, and not elmo.  That doesn't leave many options.
[01:54] <lamont> ross is d-i
[01:54] <lamont> hrm... is 23% atm
[01:54] <infinity> Yes, I noticed. :)
[01:54] <elmo> yeah, but somethings using 20% of 150Gb and it's not a package
[01:55] <elmo> and it's not d-i
[01:55] <infinity> So, there's a mystery process that eats the disk at 9am every day, and then happily cleans up after itself.
[01:55] <lamont> at 39MB, gcc-4.0 wins for logfile size
[01:55] <lamont> oh, joy.
[01:55] <lamont> 9AM DC time?
[01:55] <infinity> Yeah.
[01:55] <infinity> elmo : I would assume the majority of that 20% of 150GB is ccache.
[01:56] <infinity> We do have ungodly large caches.
[01:56] <elmo> ccache doesn't clean upitself ?
[01:56] <elmo> oh 23% USED?
[01:56] <infinity> Yes, 23% used.
[01:56] <elmo> so something is using 100Gb ... W_T_F?
[01:56] <infinity> Yeah, at 9am, something is eating 110GB.
[01:57] <infinity> Yum.
[01:57] <infinity> It is really at the same time every day, or just coincidentally rather close?
[01:58] <elmo> all between 9am and 10am
[01:58] <lamont> did tdb-dev finally show up (migrate to main)?  or did bogofilter just pick a diff machine to annoy after 9:40?
[01:59] <lamont> elmo: that's bizare!  and about 0300 for me.
[02:04] <infinity> Infinite loop in the find or checksecurity cron.daily jobs, perhaps? 
[02:05] <infinity> They clean up after themselves, and couldtake 3 hours to build a 110 gig log file.
[06:40] <infinity> Oh well, halfway there.
[09:18] <doko> lamont: hppa is lacking libdb4.3, so python cannot be installed
[02:38] <lamont> infinity: yeah, I guess it would.
[02:38] <lamont> doko: correct
[06:34] <doko> laminity: please requeue: sigc jabberoo libgnomemm2.0
[06:34] <doko> lamont, infinity: ^^^
[07:22] <lamont> doko: on ppc?
[07:23] <doko> lamont: all
[07:23] <lamont> I only have logs for sigc/ppc and jabberoo/ppc.
[07:23] <lamont> I'll kick the others in a bit, when my meeting hits a break and I can type faster...