Zotonic
Zotonic
zotonic@conference.zotonic.com
Wednesday, 20 March 2013< ^ >
arjan has set the subject to: Zotonic - the Erlang Content Management Framework
Room Configuration

GMT+1
[00:05:03] jeff.5nines leaves the room
[00:08:15] jeff.5nines joins the room
[00:25:14] jeff.5nines leaves the room
[00:26:06] jeff.5nines joins the room
[00:46:06] jeff.5nines leaves the room
[00:46:38] jeff.5nines joins the room
[07:15:35] arcusfelis joins the room
[07:19:55] Andreas Stenius joins the room
[08:56:05] aRyo joins the room
[08:56:29] aRyo leaves the room
[08:56:30] aRyo joins the room
[08:58:28] aRyo leaves the room
[09:15:39] Maas joins the room
[09:27:00] Arjan joins the room
[09:37:43] Maas leaves the room
[09:40:26] Maas joins the room
[09:56:04] Arjan leaves the room
[10:33:10] Arjan joins the room
[10:58:21] Maas leaves the room
[11:05:28] Arjan leaves the room
[11:05:31] Maas joins the room
[11:06:50] Arjan joins the room
[11:07:03] Arjan leaves the room
[11:10:42] Maas65784 joins the room
[11:10:42] Maas65784 is now known as Maas46496
[11:10:42] Maas46496 is now known as Maas84729
[11:10:42] Maas84729 leaves the room
[11:10:42] Maas84729 joins the room
[11:10:42] Maas84729 leaves the room
[11:10:42] Maas23500 joins the room
[11:10:47] Maas23500 leaves the room
[11:48:17] jeff.5nines leaves the room
[12:54:21] Maas leaves the room
[12:59:58] Maas joins the room
[13:00:29] Maas leaves the room
[13:36:33] Maas joins the room
[15:09:25] Arjan joins the room
[15:18:54] Arjan leaves the room
[15:30:58] Arjan joins the room
[15:58:58] Arjan leaves the room
[16:25:38] Maas leaves the room
[16:25:39] Maas joins the room
[17:31:33] Maas leaves the room
[17:52:11] Maas joins the room
[18:05:23] Maas leaves the room
[19:48:32] arcusfelis leaves the room
[20:01:09] Maas joins the room
[20:35:57] Arjan joins the room
[20:36:06] Maas leaves the room
[20:39:58] Maas joins the room
[20:45:54] <Marc Worrell> Anybody of the "stats people" know this error?
[20:45:55] <Marc Worrell> 2013-03-20 17:02:52.468 [error] emulator Error in process <0.32714.199> on node 'zotonic001@Lamma' with exit value: {badarg,[{ets,delete,[26940620904],[]},{folsom_sample_exdec,delete_and_rescale,4,[{file,"src/folsom_sample_exdec.erl"},{line,122}]},{folsom_sample_exdec,rescale,5,[{file,"src/folsom_sample_exdec.erl"},{line,106}]},{folsom_sample_exdec..
[20:46:04] <Marc Worrell> I get it once in a while in my dev env
[20:47:28] <Andreas Stenius> no, never seen before..
[20:47:36] <Maas> The bleeding edge..
[20:49:17] <Maas> It is trying to drop a table to clear out stats it has collected.
[20:50:25] <Andreas Stenius> but I think I switched from exdec to slide... ? is this on some customized stats?
[20:51:11] <Maas> The table was already gone probably.
[20:55:11] <Maas> That could fail in the same way too. I guess.
[20:55:25] <Andreas Stenius> ok
[20:55:38] <Maas> The process responsible for the ets tables probably died without folsom knowing about it.
[20:55:41] Andreas Stenius experimenting with wxErlang.. :)
[20:55:56] <Andreas Stenius> heh
[20:56:00] <Andreas Stenius> ouch
[20:57:02] <Maas> Looking....
[20:58:11] <Maas> That is zotonic_sup...
[20:59:51] <Maas> You would think that folsom wouldn't survive zotonic_sup crash.
[21:00:23] <Andreas Stenius> folsom isn't started under the zotonic supervision tree I suppose... ;)
[21:00:54] <Maas> No it isn't but zotonic_sup is responsible for the tables.
[21:01:16] <Andreas Stenius> doh. then folsom ought to monitor those tables, some how...
[21:04:31] <Maas> Indeed it looks like it is not part of the supervision tree. In z_stats.erl folsom:start() is called
[21:08:28] <Maas> This starts it as an app... it looks like it makes references to the ets tables.
[21:08:41] <Maas> belonging to zotonic_sup.
[21:09:25] <Maas> ets-pagetti
[21:09:35] <Andreas Stenius> hmmm...
[21:10:59] <Maas> Maybe it is just exdec?
[21:11:51] <Andreas Stenius> no idea.. :p
[21:13:16] <Maas> Looking. I thought it was logical to bind the tables to zotonic_sup...
[21:15:59] Maas leaves the room
[21:18:30] Maas joins the room
[21:20:48] <Maas> Checking how it finds the table ids in the first place.
[21:23:57] <Maas> That information is looked up in a table owned by folsom_sup in folsom app
[21:24:49] <Maas> Nice one...
[21:30:16] <Maas> Sigh
[21:34:18] <Maas> So if you call folsom_metrics:new you will be responsible for the table, but some of the bookkeeping about it is in another app. That is just weird.
[21:34:48] Arjan leaves the room
[21:45:21] <Maas> https://github.com/boundary/folsom/issues/30
[21:48:58] <Maas> I don't like the proposed fix also.
[21:53:28] <Andreas Stenius> the one from vinoski?
[22:13:03] <Maas> It is a design flaw, and the proposed doesn't fix that.
[22:13:26] <Maas> folsom should hold on to the table, and maybe monitor the process creating them
[22:15:04] <Maas> When the process dies, it can do the bookkeeping just fine. Otherwise if you fix just this part you miss out on the reverse problem when folsom dies.
[22:49:08] Maas leaves the room
[22:57:26] Maas joins the room
[23:10:24] <Maas> Gonna propose it as an alternative solution. Any table should be owned by folsom anyhow. For automatic bookkeeping folsom can monitor the creator of the metric. No dirty tricks needed.
[23:37:40] Andreas Stenius leaves the room
[23:55:18] Maas leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!