Log file Viewer

[02:17:37] <Mastema> is there any information on when mongo plans on supporting gridfs in the new c++ driver? I can't find anything on it

[04:21:31] <MacWinner> if i don't want duplicate filenames in my gridfs collection, do i just need to create a unique index on db.fs.files filename?

[04:35:14] <Boomtime> @MacWinner: that will do it, though i'm not sure how the gridfs API in various drivers will react - they create the file metadata entry first so it should be graceful

[04:35:56] <Boomtime> btw the filename is technically 'optional' according to spec, so you'll need to make the index sparse if you want that ability to be retained (some driver APIs might permit files without filenames)

[04:36:34] <MacWinner> got it.. are sparse indexes needed when using wiretiger? is the prefix compression something different?

[04:51:09] <Boomtime> ehh.. wat?

[04:51:41] <Boomtime> sparse indexes are an index condition, nothing to do with the storage engine - prefix compression is a tactic for using less space, but it does not change the functional conditions at all

[04:52:05] <Boomtime> did that explanation clear that up? (not sure if i understood the question)

[04:53:55] <MacWinner> yes, sorry, i was confused!

[04:53:56] <MacWinner> thanks!

[04:54:45] <MacWinner> also, i just cleaned out a lot of files from my gridfs (a bunch of duplicates). but the DB size didn't really go down when i did show dbs. Is this expected?

[04:57:09] <Boomtime> sadly.. yes

[04:57:26] <Boomtime> are you using wiredtiger now, or still mmap?

[04:58:37] <Boomtime> in wiredtiger there is a chance that on disk storage will be returned, but i don't think it is controllable - mmap never releases disk space for a database once assigned (unless you repair/resync)

[05:11:54] <MacWinner> Boomtime, yes, using wiredtiger

[05:12:17] <MacWinner> sorry, was in another window..

[05:13:26] <MacWinner> oh, for me I don't really care about reclaiming the space as long as it can be resused..

[06:14:01] <SaulGoodman> hello. hi. i was wondering if there was a good way to encrypt with mongodb the values that i put in there. then decrypt as i pull them out. does mongo already have libraries for that? or do i have to do that stuff manually in my server code then store that data ?

[08:57:13] <ams_> My collection is capped at 5gb but on disk I see >20gb. Any ideas why?

[10:31:01] <Avocher> Hello, i seem to have added two different kinds of spaces " " (charCode:32) and " " (charCode:8201) in my database, is there any way to configure mongodb to treat them as equal?

[10:39:16] <Derick> no

[10:53:58] <ams_> The "storageSize" and actual size on disk of my collections appear to be quite different. Any ideas why that might be?

[10:56:08] <ams_> Or size+totalIndexSize =~ 7gb, size on disk =~ 20gb

[11:03:08] <Pingula> hi can somebody recommend a resource on how to structure data within mongodb for optimal search performance??

[11:05:23] <ams_> Actually, this is more telling: https://www.irccloud.com/pastebin/tGyIWQtL/

[11:05:34] <ams_> dataSize 5gb, filesize 20gb

[11:14:09] <jokke> hi

[11:14:54] <jokke> some weird stuff is happening that makes my collection empty out every few seconds or so

[11:15:24] <jokke> how can i debug what's happening

[11:16:14] <jokke> the only write command (that i know of) that i'm using is insert.

[11:18:11] <Derick> jokke: turn on all server query/command logging and check the logs

[11:19:58] <jokke> systemLog.component.command.verbosity ?

[11:27:37] <Derick> ? you can just bumb the log level to two db.setLogLevel(2) iirc

[11:27:46] <Derick> no need to restart or change config

[11:27:52] <jokke> ah ok

[11:36:50] <jokke> ah

[11:37:45] <jokke> this is so weird

[11:37:50] <jokke> there's nothing in the logs

[11:38:00] <jokke> just the inserts

[11:38:34] <jokke> but the collection stays empty

[11:38:44] <jokke> or empties out after a while

[11:38:46] <jokke> wtf..

[11:39:14] <jokke> here's what happens according to the log: https://p.jreinert.com/mbG/

[11:40:09] <jokke> but db.messages.count() gives me 0

[11:52:44] <jokke> wtf really... i looks like my collection is periodically wiped without leaving any traces in the logs

[11:54:21] <jokke> i'm using libmongoc with crystal btw. the driver is very beta so i expect there to be bugs but i'm not sure if something like this can be caused by the driver...

[12:00:01] <jokke> here's the whole log: https://p.jreinert.com/VZNK/

[12:15:38] <Joran> I know the recommendation is xfs for mongo, has anyone done any performance testing with zfs?

[12:18:52] <jokke> zfs is cow right?

[12:19:08] <jokke> don't think that's a good thing

[12:31:22] <Joran> zfs can be set up cow

[12:32:29] <Joran> it can be set up numerous ways so it would need a fairly thorough performance test run in different configurations

[12:39:44] <jokke> yeah

[12:39:57] <jokke> but i wouldn't bother even testing with cow turned on

[12:40:44] <jokke> same applies to btrfs

[13:28:42] <Joran> hmm... looks like my info was incorrect zfs is always cow, but it is slab where btrfs is b-tree

[13:49:02] <jokke> yeah btrfs also allows you to turn cow off on inode basis

[14:23:26] <Joran> Found an interesting article : http://www.clock.co.uk/blog/mongodb-performance-on-zfs-and-linux

[22:36:06] <JasperNL> Hi :)

[22:36:20] <JasperNL> I've got a problem installing mongodb (community edition) to my Debian Wheezy box

[22:36:52] <JasperNL> I followed these steps - https://docs.mongodb.com/manual/tutorial/install-mongodb-on-debian/#install-mongodb-community-edition - but the package is not in the repo

[22:38:18] <JasperNL> The error is basically "E: Unable to locate package mongodb-org"

[22:45:20] <JasperNL> I'll try to install from tarball instead

[22:55:31] <JasperNL> Managed to install from tarball yay