pmxbot IRC Log Viewer

[03:04:20] <quattro_> i only need an arbiter if i have 2 replica's in a set?

[05:03:30] <andrewcarter> Hey guys, I have a query I want to do but I'm not sure if I'm thinking about it right

[05:03:42] <andrewcarter> say my objects have tags

[05:04:07] <andrewcarter> like tags: [1, 2, 3] in one object and tags: [1,4,5] in another

[05:04:21] <andrewcarter> and basically I want to find all objects where there are intersections of tags

[05:04:39] <andrewcarter> so like when comparing for [1,2,3], [1,4,5] would match but [6,7,8] wouldn't

[05:04:42] <andrewcarter> what's that called?

[10:35:44] <anoopcs> Cannot start mongod service on fedora 19?

[10:36:27] <anoopcs> This is the error "Starting mongod (via systemctl): Job for mongod.service failed. See 'systemctl status mongod.service' and 'journalctl -xn' for details."

[10:36:57] <anoopcs> Can anyone help me with a solution?

[10:37:32] <ranman> anoopcs how mongodb installed? from package manager?

[10:38:06] <anoopcs> ranman: via yum.

[10:39:00] <anoopcs> ranman: exactly as specified in the mongodb documentation.

[10:39:28] <ranman> it looks like there could be an error in the package at the moment

[10:39:37] <ranman> this stackoverflow solution has an answer

[10:39:38] <ranman> http://stackoverflow.com/a/19858363/240004

[10:45:23] <anoopcs> ranman : Just now tried to do so and no luck.

[12:40:01] <qswz> o :3 $setOnInsert

[13:06:29] <dawik> hey all. when using db.coll.find() can i truncate the binary members so they do not spam the terminal so much?

[16:31:17] <anoopcs> ranman : Thanks for that stackoverflow link. At first I forgot to stop the mongod service. No got it working.

[16:31:41] <anoopcs> ranman : Thanks for that stackoverflow link. At first I forgot to stop the mongod service. Now got it working.

[16:48:29] <ranman> excellent! glad you got it to work sorry I couldn't help more

[20:10:12] <jordz> Got a fairly big question!

[20:10:37] <jordz> I've got a test application running right now across 2 virtualised shards. I'm trying to work out how fast this should be.

[20:10:37] <jordz> I know it's an extremely tough question because there are many things to consider on the hardware side plus the indexing side.

[20:10:37] <jordz> Assume this, the queries I'm describing are constantly hitting the database over a period of an hour. There are no indexes and the hardware, 4GB ram, dual core.

[20:10:37] <jordz> 1 database, 1 collection. The collection right now contains 120,000 documents so far.

[20:10:39] <jordz> The test schema works as follows, it's based on weekly buckets of data which are split into days.

[20:10:41] <jordz> The days can range from having 100's of Stats per day, some have a few hundred some have a few:

[20:10:43] <jordz> {

[20:10:45] <jordz> Week: (Week Number Integer)

[20:10:47] <jordz> Name: (We use this as the Identifier for the data set)

[20:10:49] <jordz> Data: [

[20:10:51] <jordz> {

[20:10:51] <cheeser> pastebin

[20:10:53] <jordz> Date: (ISODate)

[20:10:55] <jordz> Stats: [

[20:10:57] <jordz> {

[20:10:59] <jordz> Stat Name: <string>

[20:11:03] <jordz> Number: <int>

[20:11:05] <jordz> }

[20:11:07] <jordz> ]

[20:11:09] <jordz> }

[20:11:11] <jordz> ]

[20:11:15] <jordz> }

[20:11:17] <jordz> We then have workers that kick in when the data comes in, right now I'm running 5 workers which take a chunk of data which contains 120 daily stats and inserts.

[20:11:20] <jordz> The query flows as follows:

[20:11:22] <jordz> Check if the Week Bucket exists -> (if not created it) -> check if the day xists in the week bucket (create if not) -> push the stat into the document.

[20:11:25] <jordz> I'm seeing write locks of 7 seconds, 55 queries a second and 55 updates a second but the bottle neck is completely on the database side.

[20:11:28] <jordz> The sharding key is the Name so all sets of the same name exist on the same shard.

[20:11:32] <jordz> What should I be looking for in terms of stats?

[20:11:34] <cheeser> please don't paste here

[20:11:34] <jordz> Oh crap

[20:11:36] <jordz> didn't realise notepad did that

[20:11:38] <jordz> definitely

[20:11:40] <jordz> 2 sec

[20:11:42] <jordz> sorry!

[20:11:46] <jordz> Yeah, didn't realise

[20:11:49] <jordz> won't do it again!

[20:11:50] <jordz> http://pastebin.com/eqjQxanT

[20:12:57] <jordz> Never really had to ask a question on IRC that was that big!

[21:16:58] <oskie> hello, is there a way to move a node that is PRIMARY single node in a cluster to another cluster, as secondary?

[21:17:08] <oskie> ...without destroying all data on disk...

[21:17:34] <oskie> or nevermind, i'll just remove all the data

[21:29:09] <brendan6> Hello, the documentation at http://docs.mongodb.org/manual/core/2d/ says to only use a 2d index if my database has legacy location data. Does that mean a standard index is suitable for indexing a field with GeoJSON data?

[22:07:18] <bla> Hello.

[22:07:52] <bla> Is it possible to stop mongo from binding any TCP ports? (use only Unix socket)

[22:11:39] <cheeser> i don't think so

[22:11:49] <cheeser> i can't find anything in the docs about it

[22:13:29] <bla> Yeah, I searched too.

[22:14:00] <bla> Shame a bit; I wanted to create a private mongodb instance on a server and don't really want to play/trust mongodb authentication mechanisms when unix sockets are so simple.

[22:27:21] <bla> Ok. Another one - is it possible to connect with 'mongo' to unix socket?

[22:28:23] <bla> Ok. I guess there is really no real support for unix sockets. I need to abandon this idea...

[22:34:58] <cheeser> you can use unix sockets, i thought...

[22:36:51] <algernon> indeed, you can.

[22:38:32] <algernon> not sure if you can turn off TCP, though.

[22:39:42] <bla> Ok. How to run `mongo' then? I tried mongo 'unix socket path' or mongo --host 'unix socket path' but nothing really seems to work

[22:40:06] <bla> Maybe there is some trick I haven't tried (unix://?)

[22:40:10] <algernon> mongo /path/to/unix/socket/database

[22:40:44] <bla> Sat Dec 7 23:38:43.119 starting new replica set monitor for replica set . with seed of mongodb-50000.sock:27017

[22:41:07] <bla> Sat Dec 7 23:39:18.427 In File::open(), ::open for 'mongodb-50000.sock' failed with errno:6 No such device or address

[22:41:20] <algernon> you're missing the /database part most likely.

[22:41:41] <bla> Yep! Also this needs to be a full path it seems

[22:41:46] <bla> ./file/database won't work

[22:41:48] <algernon> yep.

[22:41:50] <bla> Thanks

[22:42:18] <algernon> mighty unintuitive, and also undocumented by the looks of it. but trial and error go a long way1

Log file Viewer

Help | Karma | Search:

#mongodb logs for Saturday the 7th of December, 2013