WPWG-20080722-transcript

From GO Wiki
Jump to: navigation, search

(08:34:15 AM) ben [hitz@c-98-210-219-14.hsd1.ca.comcast.net] entered the room.
(08:34:31 AM) ben: hi
(08:35:00 AM) gwg [gwg@apple-0-17-f2-0-e5-68.dhcp.lbl.gov] entered the room.
(08:35:15 AM) ben: call in?
(08:35:24 AM) gwg: no, it's just a chat today
(08:35:34 AM) ben: kewl
(08:35:34 AM) gwg: but I don't think many others are going to show up
(08:35:35 AM) sjcarbon: it's WPWG, so it's usually been a chat
(08:36:41 AM) ben: it's been a while since we chatted
(08:38:10 AM) sjcarbon: sit back, relax, get a cup of tea...
(08:38:23 AM) Logging started. Future messages in this conversation will be logged.
(08:38:33 AM) sjcarbon: I've been looking at the load script a bit.
(08:39:03 AM) sjcarbon: I noticed that you made some changes yesterday. Should I fold them in to my debugging version?
(08:41:00 AM) gwg: hey everyone, check out the exciting new amigo URL I just created! <a href="http://tinyurl.com/amigosearch">http://tinyurl.com/amigosearch</a>
(08:41:12 AM) ben: all I changed yesterday was roll the load uniprot into a subroutine.
(08:41:27 AM) gwg: that's going to save everyone a whole lotta time!
(08:41:35 AM) sjcarbon: Ok. I'll leave mine as-is then
(08:41:38 AM) sjcarbon: .
(08:41:38 AM) ben: my feeble hope was that the sprot index was somehow not getting freed up (since it never went out of scope)
(08:43:52 AM) sjcarbon: It didn't work?
(08:43:52 AM) sjcarbon: I going to put some variable memory probey things in and see who has got all of the memory.
(08:44:10 AM) sjcarbon: It's irritating that it takes so long to get to the bug.
(08:44:43 AM) ben: yes, that's why I cry myself to sleep over it.
(08:45:01 AM) ben: I think it must be the huge dbxref2id_h hash in $apph.
(08:45:21 AM) ben: I did add some crude memory snooping code as well
(08:45:26 AM) ben: (via ps!)
(08:45:34 AM) ben: I didn't want to recompile perl
(08:45:56 AM) ben: Stuart suggested moving to 64bit Linux because the machine has like 13GB RAM
(08:46:16 AM) ben: 32bit seems to crap out around 3Gb allocated/process
(08:47:22 AM) sjcarbon: I looked at that, and couldn't find the limit written down anywhere (except for the obvious 32-bit memory mapping issues).
(08:48:34 AM) sjcarbon: Have you tried: Devel::Size?
(08:48:55 AM) ben: Memorize?
(08:49:40 AM) ben: we have another software, pathway tools, which also seems to give up at about 3Gb allocated.
(08:54:16 AM) sjcarbon: I don't think anybody is coming today.
(08:54:16 AM) sjcarbon: Anything else?
(08:54:16 AM) sjcarbon: If there is nothing else, I'm going to go back to poking the load script...
(08:55:41 AM) ben: Should I go ahead and patch amigo in production?
(08:55:56 AM) ben: Jane at least tested it...
(08:56:04 AM) ben: You guys want to bang on it for 5 minutes?
(08:56:16 AM) ben: <a href="http://goweb-dev.stanford.edu/">http://goweb-dev.stanford.edu/</a>
(08:56:36 AM) ben: I sort of have to do it today, or not for 3 weeks...
(08:56:51 AM) sjcarbon: Do you mind if I aim a bot at it until lunch?
(08:57:45 AM) ben: ok with me, let me inform sysadmins...
(08:58:05 AM) ben: drop me an email when you are "happy". I have a meeting at 1 pm though.
(08:58:22 AM) sjcarbon: no problem.
(08:59:57 AM) sjcarbon: I'll kill it just before lunch and let you know after.
(09:00:06 AM) ben: OK, that it?
(09:00:10 AM) sjcarbon: OK, I'm off...
(09:00:16 AM) sjcarbon: Later!
(09:00:22 AM) ben: I will miss any meetings until 8/11
(09:00:25 AM) ben: l8r