Log file Viewer
#mongodb logs for Wednesday the 11th of June, 2014
[00:08:36] <dexterity> I use flask and mongoengine and every time I use model_form(Task) I get AttributeError: 'dict' object has no attribute 'iteritems'
[00:09:15] <dexterity> I use it like described here: http://docs.mongodb.org/ecosystem/tutorial/write-a-tumblelog-application-with-flask-mongoengine/
[01:06:19] <Gr1> I was wondering why, dropping a database on a replica set is not reflecting on the secondary nodes
[01:26:22] <dccc> hi folks, i'd like a second opinion on my data structure. i'm building a property management applications. i have 'buildings' and within, 'units' as an array of subdocuments. the 'listings' documents duplicate data from buildings and units to facilitate search. this doesn't seem very efficient to me. perhaps sql is better for this kind of data.
[01:32:07] <Gr1> The resident memory for mongod goes down after a write happened (even it is a small write)
[01:35:03] <Gr1> Is there any way I can see if a reindexing is going on after replication, on each secondary node?
[01:56:42] <dman777_alter> I have for a field data : Number...but when I insert a string I don't get any complaints. Why is this?
[10:21:33] <rspijker> well, indirectly of course, by using the min and max shardkey values of the chunk in that command
[10:22:56] <rspijker> also have a look at this SO answer: http://stackoverflow.com/a/12470395/2156236
[10:35:41] <KamZou> I'm looking for a way to detect error (the "chunk is too big to move" one) when I do a moveChunk
[10:45:02] <ed_saturn> hello, my mongo query on a replicated sharded cluster is slow if goes to mongos (route server), if query goes to replSet instance directly, everything is fine. any clue ?
[10:47:55] <Alan> (I was trying to figure out how I'm supposed to go from a full mongodb URI to the database it refers to - answer is MongoClient(uri).get_default_database())
[10:51:41] <rspijker> KamZou: just wrap the move function with your own and add some error handling for that? If it occurs perform a split?
[12:35:19] <dragoonis_> I'm trying to back up a collection before I do some processing on it. I'm using db.collection.copyTo() but it's doing an eval underneath and LOCKING everything, so all reads are locked too. How can I back up a collection and have it not lock the mongod connection ?
[12:41:21] <dragoonis_> hey Derick I'm in PHP land right now, using the extension - I'm doing Mongo::execute("function(){ db.$from.copyTo('$to') }", array('nolock' => true));
[12:44:19] <dragoonis_> http://docs.mongodb.org/manual/reference/method/db.cloneCollection/#db.cloneCollection
[12:45:23] <Derick> dragoonis_: I think you'll just have to loop it yourself. Write a script that does a find() without query, and write each entry to the new collection
[13:21:55] <cozby> mongo documentation talks about snapping EBS often (on AWS of course), but snapping an EBS requires that disk be in a static state, and no writes are taking place
[13:22:37] <cozby> so I'm guessing that one must remove a replica instance from a replica set, then snapshot the ebs volume, then re-attach
[13:25:09] <Derick> cozby: http://docs.mongodb.org/ecosystem/tutorial/backup-and-restore-mongodb-on-amazon-ec2/ should have everything you need
[13:30:28] <cozby> here's what I'm trying to do (let me know if its silly) - Have my replica set in an autscaling group, so if one instances goes down a new one comes up, this new instance startup will pull the mongo files from a mongodump I take daily and store on an s3. The new instance will copy the dump data and will be able to sync with the replica set again.
[13:31:41] <cozby> if a new instance is launched the data will be too stale, hence me copying the daily data dump at boot
[14:30:54] <Joeskyyy> Ah, you mean as the backend store. I can't say actually, something I don't control
[14:54:59] <Derick> for id in ids { collection.update({_id: id, count: old_count[id]}, { '$set': { count: count[id]}) } }
[14:56:15] <saml> the app queries like this: var date = coll.find().sort({date:-1}).limit(1)[0].date; coll.find({date: date})
[14:56:50] <saml> and the update script is doing coll.update({_id:id}, {$set:{count:count[id], date: new_date}})
[14:57:32] <saml> the loop doesn't take too long to run.. if i refactor the script so that there's only one loop
[14:57:48] <Derick> for id in ids { collection.update({_id: id, date: old_date}, { '$set': { count: count[id], date: new_date}) } }
[14:58:00] <saml> but actually, it's talking to omniture. get the counts. do the loop. talk to different service. do different loop.. so the whole script takes 10 minuets to run
[15:00:06] <Derick> but by checking for the old_date, you can run it multiple times without problems
[15:00:18] <saml> it's not really mongodb. i just need to refactor the script so that it collects counts from various services. and do the update in one loop
[15:02:01] <saml> let's say script updated one document. new_date is updated. so the app tries to find({date: new_date}), which returns single document only
[15:04:02] <saml> instead of finding the lastest "date" in coll and do coll.find(date == latest_date_found)
[15:36:58] <Shakyj> Hey, I am thinking on storing google ranking data in MongoDB, just trying to think of my structure, access will be mainly keyword then domain based so I came up with this structure https://gist.github.com/anonymous/60ed1fef3184b5134226 but would that allow me to sort the rankings by date? If so how?
[15:38:49] <dragoonis_> When I run it against 'admin' I get: string(47) "invalid collection name: answer_to_response_map"
[15:39:30] <dragoonis_> 'answer_to_response_map' is the TO part, so of course it doesn't exist, I'm trying to rename it to that
[15:46:43] <Shakyj> Thanks Derick I will just now have to pour over the docs to find how to best query it :D
[15:48:09] <dragoonis_> return $this->getMongoAdminDatabase()->command(array('renameCollection' => $from, 'to' => "$dbName.$to"));
[15:48:35] <joshua> dragoonis_: http://docs.mongodb.org/manual/tutorial/manage-the-database-profiler/
[18:24:58] <ForSpareParts> Is there a way to retrieve only part of a Mongo document? I anticipate specific pieces of my documents getting huge, but it’s important I be able to get their top-level attributes without pulling down loads of data every time.
[18:25:11] <ForSpareParts> I’m using Mongoose, and I’m a total noob. Sorry if this is a stupid question.
[18:28:33] <cheeser> ForSpareParts: http://docs.mongodb.org/manual/reference/method/db.collection.find/#projections
[19:02:16] <kali> gsd: une query will probably be faster (because you'll have much less network overhead)
[19:08:29] <kali> cppiii: i think you're confused somewhere. mongodb and mysql have totally different data models
[19:12:13] <cppiii> kali: yes, but what I want is reading the updated data from mongodb and insert them into mysql with different schemes
[19:21:02] <saml> coll.find({date:date}).sort({count:-1}).limit(10); vs. coll.find().sort({date:-1,count:-1}).limit(10); what's better?
[19:23:12] <saml> find({date: mostRecentDate}).sort({count:-1}) vs. find().sort({date:-1, count:-1})
[19:35:52] <kali> saml: provided that you have a composite index on date and count, it should be more or less equivalent
[20:05:26] <NaN> why does $unset update to null? ex: $unset: { array.pos: "" } >>>> turns null in pos
[20:06:56] <cheeser> what are you expecting to happen there? that that position would vanish entirely?
[20:45:25] <diphtherial> i'm not sure if this is a commonly-asked question, but just to verify there isn't guarantee of atomicity when updating multiple documents, right?
[20:47:01] <diphtherial> i was reading a bit about tokumx, which apparently offers transactions; what's the opinion about it here?
[20:47:18] <diphtherial> (the transactions unfortunately aren't going to work for us, since we're using mongodb with shards)
[20:59:17] <kali> diphtherial: i'm messing around with an alternative approach, if you want to have a look and comment... https://github.com/kali/extropy
[21:21:40] <ForSpareParts> How do you represent an arbitrary dictionary in a Mongoose schema? Something where the keys don’t have any special meaning, and they’re just used as an index to a model.
[21:23:28] <ForSpareParts> In schema terms, there’s nothing special about the names ‘somewidget’ and ‘otherwidget’, they’re just how widgets are identified. But I don’t see a structure like that in Mongoose, so I’m assuming I should be doing something else — maybe making the name/id part of the widget somehow, and representing “widgets” as a list of objects...?
[22:57:40] <Jadenn> hiya. so i'm having an issue with sorting, i need to retrieve the newest records, which is fine, but it's not letting me sort again so they are in the right order.
[23:02:56] <Guest92688> any one already got a situation where the mongos stop no answer mid connection on connections to localhost duo to load?
[23:05:52] <NaN> cheeser: still couldn't do $pull with index. ex: { foo: [{},{},{}] } < I would like to remove the 3rd {} something like $pull: { foo.3: true } ...any clue?
[23:24:45] <b1nd3r> Hello. If my docs have boxes recorded inside a field, how can I search in mongo which boxes (docs) a point belongs too?
[23:31:06] <b1nd3r> NaN: a box, with a bottom left coordinates and an upper right coordinates inside my "bounds" field and type "Box". They represent a box in a map, how can I take a point and search in my database which boxes this point is inside of?
[23:34:27] <b1nd3r> NaN: if I record points I can use $geoWithin : { $box ... but in this case is the inverse