Log file Viewer
#mongodb logs for Friday the 11th of July, 2014
[01:17:41] <danfinch> is it true that mongorestore with --collection and --drop will only drop the collection specified in the command line?
[01:23:44] <danfinch> and if I do not provide collection, does it only drop those which already exist at the destination?
[01:24:29] <cheeser> import into a new db. create a new collection or two, then import with --drop and see
[01:28:40] <danfinch> will do, would like to see if someone who knows they know knows, too, since I can't find anything that clear in docs/basic googling
[06:53:59] <samurai2> hi there, anyone know how to access field with space with dot notation in mongodb terminal? thanks
[06:54:37] <samurai2> so for example there is mongodb document looks like this { "first name" : "alpha"}
[07:09:36] <shambat> I have a python script that loads in two large json files, this is causing my VPS to crash since all the memory is being used up. I would like to store these files in a database like mongodb instead. Is there a way to get mongodb to use disk instead of memory, or set a maximum or something? I am not as concerned with speed.
[08:36:37] <sweb> i have mirror collection ... latest_news and archive_news i want have all latest news on archive_news
[08:43:22] <rspijker> inside of a single collection it needs to be unique, yes. But if you have two documents, say a and b from collections A and B respectively. Then there is no objection to a._id == b._id
[09:45:04] <user123321> If I could open a mongo-db using robotmongo which has been hosted remotely, how to get that database up and running in my Ubuntu/LUbuntu host?
[10:31:06] <user123321> rspijker, how to get a copy of the mongo-db database which is hosted in mongolab.com? I can access it using robomongo. I'd like to host in my machine itself.
[10:54:47] <kas84> I’ve just installed mongo-10gen-server on my aws machine but don’t see the “mongo” command
[10:56:38] <joannac> http://docs.mongodb.org/v2.2/tutorial/install-mongodb-on-red-hat-centos-or-fedora-linux/#packages
[12:16:48] <similian> should i be using mongo or mysql for the evaluation of apache access logs ( page visitors, pageview, referrer by unique visitor, aggregate unique visitors ... )
[12:28:11] <lfamorim> Hello! Someone knows why this query took so long to run on Mongo 2.6? http://pastebin.com/1CdWtHQF
[13:47:23] <_rgn> if I embed "user" document originally in users coll, to another document, what should the id field be called in the subdocument?
[13:59:42] <Breaking_Pitt> Hi guys... I have mongo 2.4.10 and I'm getting this error running db.stats();
[14:02:22] <oceanx> hi, I wrote an aggregation pipeline, and the last step is a projection, I want to return only the content of one of the fields (which is a dictionary) is there a way to do so without having to return the full document or projecting every single field of the sub-document?
[14:07:37] <cheeser> i'm not sure i know the answer but it'd help if you pastebinned somewhere what you tried and what you got back.
[14:14:47] <nacer> what do you thin about a shard with member with different memory size ? like node1 -> 64Go , node2 -> 64Go, node3 -> 16Go
[14:17:13] <adamcom> you can just set it to priority 0 to make sure it never becomes primary - I'd consider making it a hidden node also, that way you don't hit it for queries to the replica set (but you could send it queries directly)
[14:17:37] <Derick> > db.foo.aggregate( [ { $project: { _id: 0, amsterdam: { $hour: [ ISODate("2014-07-10T14:05:15"), "Europe/Amsterdam" ] }, tokyo: { $hour: [ ISODate("2014-07-10T14:05:15"), "Asia/Tokyo" ] } } } ] );
[14:19:01] <adamcom> nacer: if you keep traffic away from it that might hurt it with the lesser RAM, and have it at priority 0, then you should be fine
[14:19:08] <nacer> adamcom: i want to know if someone is going to says "WARNING BAD IDEA" because ...
[14:19:41] <adamcom> nacer: as long as you know the implications - the real danger is that a lesser spec machine can't keep up
[14:28:52] <oceanx> @cheeser sorry but I can't paste right now cause there's an NDA on data, but I solved just denormalizing data
[15:14:16] <San1ty> Hi Guys, I´m stuck with a rather annoying issue that I can fix. I have two processes connecting to the same mongodb database but writing to two separate collectiosn. the process write to their respective collection every 10 seconds.
[15:14:28] <San1ty> Now at sporadic moments both my process exit at the same time with the following error
[15:14:29] <San1ty> MongoError: E11000 duplicate key error index: mydb.mycoll.$_id_ dup key: { : ObjectId('53b8b6c8292b875b0a31914f') }
[15:19:23] <San1ty> and itś entirely possible the process runs fine for 5 days and then suddenly it´ll error out it´s very sporadic
[15:21:30] <_rgn> if I embed "user" document originally in users coll, to another document, what is the ObjectId typically called in the subdocument, simply _id?
[15:22:38] <rspijker> _rgn: unless you have a good reason to call it something else, that shouldn;t be a problem
[15:24:19] <San1ty> rspijker, I´doing a bulk insert of a couple of entries collection.insert(entries, function(err) {
[15:26:13] <San1ty> rspijker I let the ID be autogenerated and the rest are just a bunch of numbers with one field being a timestamp
[15:26:51] <San1ty> rspijker, i´m using mongojs as a rapper for the mongoDB API https://github.com/mafintosh/mongojs
[15:27:38] <San1ty> rspijker, is there some kind of rule against inserting at the exact same time to two different collections
[15:32:52] <rspijker> well… you could generate an ObjectId in the driver instead of having mongod do it
[15:33:12] <rspijker> apparently this guy had a similar issue: http://stackoverflow.com/questions/16133602/during-insert-e11000-duplicate-key-error-index
[15:35:02] <rspijker> but I’m not familiar with the node driver, so only going off of that SO question
[15:37:55] <rspijker> Only thing I could think of is that mongo thinks the just deleted _ids are now available again, but they’re still in the index or something… Moving the id creation to the driver should sort that. But it’s a stretch anyway
[15:42:13] <user55> Hello! I would like to ask a question, about finding and matching embedded arrays, array items.
[15:42:23] <user55> http://docs.mongodb.org/manual/tutorial/model-tree-structures-with-ancestors-array/
[15:42:26] <user55> This solution would be the best for us, our client-side applications works with the same approach.
[15:42:33] <user55> A document may look like this: {path:['path', 'to', 'resource', 'one'], resource: 'I am a Resource!'}
[15:43:33] <user55> path/to/resource/one: matches a single resource, this is the most important for us
[15:43:44] <San1ty> rspijker, might the reason be that I insert in the callback of the delete query?
[15:44:34] <rspijker> San1ty: it could be… but then that would be driver specific, I think… again, no clue on how node driver works :(
[15:47:13] <rspijker> easiest solution to that I can think of is making your ancestor array an array of documents instead of strings. Where each document has the name of the ancestor and the level
[15:48:46] <rspijker> your example: {path:[{n: path, l:5},{n:to, l:4},{n:resource, l:3},{n:one, l:2}], resource: “I am a resource”}
[15:49:10] <rspijker> in this case I made the levels relative to the resource, that might not be the best option
[15:49:56] <kali> it might be enough to store the position (l) only once, as the position in the immediate parent
[15:50:33] <rspijker> I’m assuming you;ve also considered this model: http://docs.mongodb.org/manual/tutorial/model-tree-structures-with-materialized-paths/
[15:50:46] <kali> so you store the path array (provide breadcrumbs and subtree search) and the position in the parent in each doc
[15:51:51] <rspijker> the path array approach does impose some limitations… you can’t have the same resource name in different locations without breaking things
[15:53:43] <user55> what about this: {path:[{1:path},{2:to},{3:resource}], resource: “I am a resource”}; is this valid?
[15:56:21] <rspijker> user55: so, what was the reason for not choosing that materialized path approach?
[15:57:37] <rspijker> the two queries you find important can be optimized with an index. And the ones you find less important can be supported as well (though not as fast)
[16:05:48] <lfamorim> Hello! Someone knows why this query took so long to run on Mongo 2.6? http://pastebin.com/1CdWtHQF https://stackoverflow.com/questions/24697892/slow-indexed-query-whats-wrong?noredirect=1#comment38300178_24697892
[17:58:39] <lfamorim> Someone know why this query took this long to run? http://pastebin.com/h8auApjS 1,287 sec
[18:02:08] <zhodge> I have a db of logs and right now a redundantly named collection called logs which stores log documents that all have a `type` property distinguishing certain logs from others
[18:02:20] <zhodge> would it make more sense to simply have those distinct type values as collections?
[18:08:53] <cozby> how does one purge all user info? Even after deleting all the data in my /data dir
[18:10:23] <jchamberlain> if you've deleted everything but auth is still required, wouldn't that be expected? (mongo noob here)
[18:10:51] <cozby> well yeah, auth is required, but if you launch a fresh mongo instance with auth connecting without auth works fine
[18:21:19] <jchamberlain> in PHP, is it possible to return a query result as a json string instead of an array?
[18:36:57] <sebastian_o|> hi guys, what you consider good API design when your API needs to answer a create when the object already exists? bad request? server error? or a 200 with a false in payload?
[19:02:16] <lpin> you could take a look at what popular APIs return for conflicting resources like github, twitter etc.
[19:03:40] <sebastian_o|> "Git DB API functions will return a 409 Conflict if the git repository for a Repository is empty or unavailable. This typically means it is being created still. Contact Support if this response status persists."
[19:22:14] <lpin> sebastian_o|, from your link it seems that github return 422 with a "validation failed" message and a "already_exists" code
[19:41:27] <luis_alen> Can a single mongo instance act as arbiter for different clusters/replica sets?
[21:44:32] <styles_> I am doing it with map reduce right now, but I'd like to try the aggregate framework if possible
[23:38:57] <travisgriggs> i have a test/prototype mongodb on a 32 bit linux install. i have a new 64 bit linux machine running mongo now… is there an easy way to copy the data from the one to the other. i’m getting createPrivateMap errors on the 32 bit machine, so it seemed time to remove the 32bit as a potential problem
[23:40:02] <travisgriggs> i assume it’s not as simple as copy the /var/lib/mongodb directory (that would be the cat’s meow if it WAS that easy…)
[23:55:00] <travisgriggs> ok, i found mongodump, but i can’t even get it to run. i get the open/create failed to createPrivateMap error. anyone know how i can diagnose/fix that error?