=== stub [n=stub@ppp-58.8.2.53.revip2.asianet.co.th] has joined #launchpad-meeting [09:59] https://launchpad.canonical.com/PrivateBranches === ..[topic/#launchpad-meeting:SteveA] : public launchpad meeting channel [10:03] Waaaay offtopic [10:04] Porno name for 'Launchpad' [10:04] Raunchpad? [10:04] Raucnh Pad [10:04] need the second syllable [10:05] Blewpint.launchpad.net [10:12] I just lost my connection. Could you invite me again? [10:14] SteveA: ^^ ? [10:15] ok [10:15] thanks [10:19] select * from pg_locks where pid = pg_backend_pid(); [10:21] SteveA: that was fast [10:26] SteveA: dyson needs some small love. Is it possible that the infrastructure team can do that ? [10:27] https://launchpad.net/products/launchpad/+bug/51168 [10:37] james has agreed to do this [10:37] stub mentioned that dyson should be made more robust [10:39] agreed and agreed. thanks jamesh [10:48] stub: on pagetests... [10:49] i wonder if we can do a GET vs POST thing, and use a r/o connection for GETs [10:49] then we could always rollback in the publisher on a GET [10:50] and then not need to do fancy shit when pagetesting [10:50] Thats a thought [10:51] need to be a bit careful about not rolling back session changes [10:51] Hmm... but we want to make GET requests use a read only database connection anyway, so we know no changes are made using that. [10:51] yes [10:51] but i wouldn't want to rely on that convention for the pagetests [10:52] i would prefer to rely on "no commits have been made" [10:52] although i suppose session stuff screws that a bit [10:52] although resetting the session DB should be cheaper than resetting the main DB [10:52] We don't care about updates to the session database [10:52] if you insist [10:52] sounds a tad dodgy [10:52] we just don't use sessions much in pagetests [10:53] it's almost all basic auth [10:53] We do for cookie auth and notifications, many of which are tested for. [10:53] However, I decided that we will never worry about resetting the session database in the tests because if you care, you are using sessions incorrectly. [10:54] ok. i hope that assumption exists somewhere more concrete than this irc channel and your head [10:55] I think there are comments to that effect in the guts of the database reset code ;) === stub doesn't check, just in case ;) [10:58] i'd expect some garish ascii-art on server start-up [11:01] SteveA: be careful what you wish for... [11:09] aacows [11:13] stub, spiv, jamesh: I mailed a call summary to the list. You have Action Items. [11:14] http://www.fatalexception.org/action_item.html