IRC log for #koha, 2006-12-23

All times shown according to UTC.

Time S Nick Message
12:02 paul kados around ?
12:02 paul I'm at SANOP today. and we face zebra problems, that appear/disappear without any understandable reason.
12:03 paul a question : do you have shadow=ON ?
12:03 paul (because we have, and we investigate if it could be the culprit...)
12:03 paul we don't update zebra with koha-ZOOM, but with the zebra_update script from tumer, that uses ZOOM, but runs in a crontab
12:42 kados hi paul
12:42 kados yes, I have shadow on
12:42 paul ok, thanks.
12:42 paul it's really strange, and probably related to mod_perl then...
12:43 kados what is happening?
12:43 paul since this morning (everybody working on Koha, 100 computers), we had 4-5 desktops that had a temporary failure
12:44 paul searches don't work for some minuts on this desktop only.
12:44 paul (2-3 minuts)
12:44 paul all other desktops working correctly
12:44 paul don't work means : searching returns "no answer to your query"
12:45 paul not a internal server error or something weird. Just as if they searched an unexisting term
12:48 kados hmmm
12:48 kados have you checked the zebra logs for errors?
12:48 kados just one desktop? that's very strange!
12:55 paul kados : the query arrives at zebra.
12:56 paul & it was once on 1 desktop, once on 1 also, and once on 3.
12:56 paul it lasted 2-3mn, then, things where OK again.
12:56 paul NO restart of zebra
12:56 paul no restart of httpd twice, a restart of httpd once.
12:56 paul a very strange problem...
12:56 kados hmmm
12:57 kados I can't imagine how it could happen
12:57 kados as http is stateless
12:57 paul maybe it has something to do with mod_perl
12:57 paul as mod_perl has persistant connexions between zebra & httpd
12:57 kados ahh, that's true
12:58 kados hmmm
12:58 kados so maybe it's caused by asynchronous searching using the same connection?
12:58 paul + httpd perl processes have the "time to live" parameter (=how many http request they serve before dying)
12:58 kados you could perhaps create a test case script
12:59 kados use fork() to simulate
12:59 paul good idea.
13:12 thd kados: are you there?
13:13 kados thd: yep
13:14 thd what does paul mean that you had found problematic for large collections in Zebra 1.X which may continue in Zebra 2.X
13:14 thd ?
13:15 thd kados: I refer to his recent message on the koha-devel list
13:18 thd kados: I had missed your reply to paul on koha-devel list previously but I see it now
13:19 thd you seem not to identify the problem paul had originally subscribed to your report
13:20 thd s/subscribed/ascribed/
13:23 kados there is a problem
13:23 kados but it's not fatal
14:00 saorge_ hi
14:00 saorge_ i'm just finishing to install koha on my laptop
14:01 saorge_ but I can connect on the http://localhost:8888/cgi-bin/koha/ page
14:01 saorge_ can you tell me how I can create a new user ?
14:11 hdl saorge: First user is koha db user.
14:11 hdl and password.
14:11 hdl Then you can add som others with members.
14:12 kados paul_: any progress on the zebra probs?
14:15 saorge hdl: thanks
14:16 hdl paul_: not around.
14:16 hdl And it is a BIG slow.
14:16 hdl saorge: you're welcom
14:21 phoenixh hi :)
14:35 kados I have 4 dev_week installs that are all very very fast for circ
14:35 kados much faster than rel_3
14:35 kados rel2 I mean
14:35 kados I wonder if it's because of innodb
05:36 hdl bonjour
05:58 js hello

| Channels | #koha index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary