2008-05-04T01:51:38 *** dbs has quit IRC 2008-05-04T05:39:49 *** ibbo has joined #OpenILS-Evergreen 2008-05-04T06:37:01 *** ibbo has quit IRC 2008-05-04T13:07:17 *** dbs has joined #openils-evergreen 2008-05-04T13:43:47 miker_: heh - we might want to warn folks about the effects that the zzz_update_materialized_simple_record_tgr trigger has on bulk loads 2008-05-04T13:44:20 * dbs was puzzleing over why his load was taking > 1000 minutes and finally put 2 and 2 together 2008-05-04T14:43:37 I guess I should be a better developer and backport my patches to rel_1_2 2008-05-04T15:00:29 miker_: for the upgrade script, methinks that we want another COMMIT / BEGIN around line 93 2008-05-04T15:01:57 actually, insert the COMMIT; BEGIN; at line 120, should do the trick 2008-05-04T15:24:18 may also need the use_staged_search element in open-ils.search definition in opensrf.xml 2008-05-04T17:44:33 dbs: see the instructions for 1.2 upgrading -- they already include the addition of use_staged_search 2008-05-04T17:45:40 and ... why do we need another commit/begin? 2008-05-04T17:51:49 dbs: or did you mean use_staged_search in the example config? 2008-05-04T18:06:25 yeah - meant in the example config 2008-05-04T18:08:24 miker_: also - extra commit / begin required because DROP VIEW reporter.overdue_reports; on line 93 triggers an error, preventing everything following it from being committed. 2008-05-04T18:09:01 food time 2008-05-04T18:43:21 dbs: I see .. I'll move them out of the transaction with the other DROP 2008-05-04T19:34:25 miker_: would it make sense for staged search to break transcendence? 2008-05-04T19:39:48 no, that should still work 2008-05-04T19:40:50 hrm 2008-05-04T19:41:36 I have 350K records loaded at http://osls.cs.upei.ca/ with source = 3; before the 1.2.1.4->1.2.2.0 upgrade the records were visible 2008-05-04T19:41:54 but now we're back to just the six manually added ones 2008-05-04T19:43:11 I could make staged_search false and see if that makes a difference 2008-05-04T19:44:15 that'd be a good first test 2008-05-04T19:46:41 line 533 of the current rel_1_2 upgrade script is the transcendant (yes, it's misspelled in the bib_source table too) 2008-05-04T19:47:25 yep - it's tied to staged search 2008-05-04T19:47:46 hrm... that's one of the things I tested very carefully on dev... 2008-05-04T19:47:51 just set it false, restarted perl services, and tried a new search with great success 2008-05-04T19:47:59 hrm indeed. 2008-05-04T19:48:09 I'll turn it back on and try again 2008-05-04T19:53:41 yep - turn it back on, and get zero results outside of the six items 2008-05-04T19:55:25 hrm... yep 2008-05-04T19:59:17 mv 1.2.2.0 1.2.2.0-rc1 2008-05-04T19:59:18 :) 2008-05-04T20:04:28 ha 2008-05-04T20:08:20 btw, kevin is probably going to come in tomorrow to set up his workstation and learn about installing opensrf and evergreen 2008-05-04T20:08:59 good 2008-05-04T20:09:24 bedtime for amber - bbiab 2008-05-04T20:09:49 um... 2008-05-04T20:09:51 huh 2008-05-04T20:09:58 (oh - and about glacial load times - that trigger is the culprit, right?) 2008-05-04T20:09:58 now it worked 2008-05-04T20:10:05 woah, weird 2008-05-04T20:10:12 on dev.gapines.org? 2008-05-04T20:10:16 yep 2008-05-04T20:10:21 http://dev.gapines.org/opac/en-US/skin/default/xml/rresult.xml?rt=keyword&tp=keyword&t=shakespeare%20king%20lear%20marlowe%20&l=7&d=2&f=&av= 2008-05-04T20:10:25 ghosts! 2008-05-04T20:11:03 well, I uncommented some RAISE NOTICE things .. but those happen /after/ the transcendant test 2008-05-04T20:12:44 commented them back out, added -foo (to avoid caching in memcache) and it's happy 2008-05-04T20:13:02 so... I'm going to call it memcache... 2008-05-04T21:02:09 ok 2008-05-04T21:16:25 but I just stopped apache, stopped all services, restarted memcached, started all services, and started apache - and no love 2008-05-04T21:16:40 wth... 2008-05-04T21:24:40 dbs: man ... I'm confused, because I just tested on demo (staged search that's in 1.2 svn) and it works 2008-05-04T21:24:58 and the upgrade script was created with cat >> 2008-05-04T21:25:17 so a couple of variables: 2008-05-04T21:25:50 I'm running on postgres 8.1 with 1.5 GB of RAM - but there's nothing in the logs indicating hitting limits or errors 2008-05-04T21:26:35 well, I'd recommend 8.2 right off the bat ... 2008-05-04T21:26:39 was demo upgraded via the upgrade script? 2008-05-04T21:27:08 miker_: your pervy love of 8.2 is indeed well known 2008-05-04T21:27:12 no, it was updated via the 300.foobar.sql ... but like I said, same code via cat >> 2008-05-04T21:27:34 dbs: you'll need up be on 8.2 for 2.0 anyway, fwiw 2008-05-04T21:28:37 fwiw I do have 8.2 source sitting ready to be configured + compiled - just balked at introducing yet another set of variables 2008-05-04T21:38:27 guess I could take a stab at etch-backports. hmm. 2008-05-04T21:49:24 *** miker_ has quit IRC 2008-05-04T21:58:31 *** miker_ has joined #openils-evergreen 2008-05-04T22:11:18 dbs: would you mind loading 300.schema.staged_search.sql into your db directly, and then try your search again ... but add -fooobar to avoid cached results? 2008-05-04T22:26:50 *** kbeswick has joined #openils-evergreen