PMXBOT Log file Viewer

Help | Karma | Search:

#mongodb logs for Thursday the 1st of September, 2016

(Back to #mongodb overview) (Back to channel listing) (Animate logs)
[10:04:37] <kevc> does anyone know how to kill off a shard which is stuck trying to move non-existant chunks?
[10:07:46] <kevc> balancer is looping over and over with - moveChunk result: { ok: 0.0, errmsg: "ns not found, should be impossible" }
[13:52:46] <Guest24> So what's the fastest, not necessarily most accurate, way of getting a collection count
[13:54:13] <StephenLynx> pre-aggregation.
[14:48:53] <Twinner> Is it possible to do a regex capture with a mongo query?
[14:49:07] <Derick> I think you can only use it for matches
[14:49:10] <Twinner> Or in an aggregation
[14:49:22] <Twinner> thats unfortunate
[15:12:18] <Guest24> If sometimes I need to sort a huge collection by a date by DESC, and sometimes ASC - Should I make an index for both? Or just one
[15:13:19] <StephenLynx> just one.
[15:13:29] <StephenLynx> mongo is able to use indexes in reverse order.
[15:16:33] <Guest24> Ok perfect, thanks!
[17:39:46] <elauqsap> hello, I am having a little trouble creating a unique index and I was wondering if anyone could help? I am using robomongo and mgo (the community golang lib) to create the indexes
[18:18:23] <GothAlice> https://jira.mongodb.org/browse/SERVER-25717?focusedCommentId=1374828#comment-1374828 — I was too bored to wait for upstream. ;P
[18:58:07] <GothAlice> Patch compiling now.
[18:59:17] <r4z> Hello there. I'm using a mongodb-input-plugin on logstash but when I reload elasticsearch with the new conf I get this:
[18:59:24] <r4z> "WARNING: Failed to load native protocols db"
[18:59:52] <r4z> Does anyone knows how to fix it? I'm running ELK on a docker container and I can't be certain if it's a java error or docker error or ELK error...
[19:23:20] <obiwahn> -
[19:56:27] <GothAlice> https://github.com/mongodb/mongo/pull/1112 ← dun dun daaaaah!
[19:57:21] <r4z> wooot
[19:57:35] <GothAlice> I'm, like, a contributor or something now. Not just a pretty support face. ^_^
[19:57:56] <cheeser> it never helps to hurt.
[20:03:07] <GothAlice> cheeser: Even compiled the first time, though I had the sign wrong when using it (subtracting instead of adding the negative index). T'was silly, but a little coffee fixed that quickly.
[22:12:58] <wgreenberg> hi, is there any way to do an atomic operation that looks like "update document A if document B.foo == bar"?
[22:14:23] <wgreenberg> I'm having an issue w/ race conditions where if I do "find(B).then((B) => if (B.foo == bar) update(A))", then B.foo could be changed between the end of find and when update happens
[22:24:43] <GothAlice> wgreenberg: No, there are no operations cross-document.
[22:25:28] <GothAlice> In order to do that, you're very likely going to need to investigate locking or two-phase commits. https://docs.mongodb.com/manual/tutorial/perform-two-phase-commits/
[22:28:04] <GothAlice> I.e. Select for and lock B against changes to that value (update with query, checking the # updated count to confirm it worked), then update A, then unlock B. Other updates to the B.foo should then be conditional on that lock not being present.
[22:28:49] <GothAlice> This introduces a) a potential delay or retry condition for updates to B.foo, and b) introduces a possibility for hanging locks, but would prevent the A update race condition.
[22:28:53] <GothAlice> wgreenberg: ^
[22:29:12] <wgreenberg> GothAlice, unfortunate but thanks, I'll look into that!