Log file Viewer
#mongodb logs for Thursday the 7th of November, 2013
[00:52:58] <lastk> hi guys, i have this object: {'title': "Hello", 'posts': [{'message':'messages'}]} how can i do a update to add a new object to posts ?
[02:12:01] <neeky> azathoth99, you create a replica if you want to backup online, or you dump the database if you can afford to lock it while the dump happens
[02:14:14] <neeky> anyone want to talk schema validation design patterns? I'm trying to decide how to design a schema that will allow user/customer defined validations
[02:15:00] <neeky> i'm currently using mongoose, but use about 1% of what mongoose does, and have about decided I don't need it
[03:10:26] <tpayne> is it a bad idea to use a records _id value as a file name that's associated with it?
[03:43:21] <baegle> I am writing a script to dedupe some data. I have identified 2 records and I would like to strip their IDs and then create a checksum of their string representation. Do I have functions like stringify() and md5() to help me here?
[06:31:34] <k_sze[work]> Out of curiosity: what happens if there is a netsplit among the servers of a replica set?
[06:32:29] <k_sze[work]> e.g. 4 replicat set members: A, B, C, and D. A and B can still talk with each other, C and D can still talk with each other.
[06:49:50] <svector> ron, I have a collection by the name dudes and I'm trying to do crudb.dudes.find()
[08:53:06] <durre> I'm going to write a newsfeed using mongodb. right now I've decided on "fan out on write". so I write the same object to all users that should see that event. then have an index on userId. how will that scale if there are millions and millions of documents? by doing this way, updating the document is not a good idea I guess?
[09:15:00] <durre> NodeX: sure. so we have the concept of followers & followees. potentially a user can have _lots_ of followers. Lets say 100 000. when a user performs certain actions the followers should see that. its important that the read of the newsfeed is quick. writing isnt as important but shouldnt be terribly slow either. so how should I model it
[09:21:14] <durre> that's more of a "write bottleneck", which is hidden from the user. I basiclly only use the followers to determine who to write the newsfeed events to
[09:22:51] <NodeX> it's a read bottlenek when you need to read the followers the user has following him to fan out to them
[09:25:08] <durre> but the user never reads the followers. its a background job that reads them and then writes the newsfeed events
[09:29:53] <k_sze[work]> I'm confused. So the official `mongo` command line client can automatically "detect" and "resolve" the replica set by giving it just one of the servers?
[09:30:40] <k_sze[work]> whereas client libraries like PyMongo need to be specifically told the addresses of all members in the replica set?
[09:31:07] <kali> k_sze[work]: it's just good practise to give all of them when possible, in case the one you're passing is down
[09:33:05] <kali> k_sze[work]: to answer your previous question, an even number of replica in a set is a bad idea, because it allows 50%/50% splits, in case nobody get a majority
[09:34:44] <k_sze[work]> So what's the proper way to specify multiple addresses with the `mongo` command line client?
[09:37:34] <k_sze[work]> And how does it tell the difference between a simple replica set and a cluster? (Somehow the pymongo library makes a distinction, with separate classes for MongoClient and MongoReplicaSetClient)
[09:38:03] <k_sze[work]> (but MongoClient is used for both single host and clusters, which is doubly weird.)
[09:39:14] <k_sze[work]> In terms of hierarchy, one would think it goes from single-host -> replica set -> cluster
[09:45:06] <k_sze[work]> Even the latest doc at http://docs.mongodb.org/manual/reference/program/mongo/ doesn't say anything about replica sets or clusters.
[11:58:45] <liquid-silence> the client is posting base64 in the body and I want to save that file into gridfs
[12:01:31] <Derick> why? not sure which language you're using, but the PHP one allows to read from a variable to store it into gridfs
[13:17:25] <bmcgee> hey if you have an object { foo: 1, bar: 2} being use as the _id for a document, is it possible to sort by _id.foo for example?
[15:02:04] <liquid-silence> db.items.find({ "item.subarray._id": "id"}, {$set: { subarrayitem.deleted: true }}
[15:14:28] <liquid-silence> Nodex is there a way I can exclude items that have been flagged as deleted when I do db.items.find({ _id: "myid"})
[15:14:42] <liquid-silence> its currently bringing back the deleted subarrayitems and the non deleted ones
[15:28:56] <liquid-silence> say I have document { name: "test", array { item1: { deleted:true }, item2: deleted:false}
[15:29:10] <liquid-silence> if I fetch the document I would want to exclude the deleted items in the aray
[15:37:46] <Nodex> liquid-silence : the docs wont help you with it, the docs refer to removeing (deleting) items
[15:38:31] <liquid-silence> Nodex I can do this, db.collection("products").findOne({_id: id, deleted: false}, { books: { $elemMatch: { deleted:false } } }
[15:42:24] <Nodex> liquid-silence : I don't really have a clue what you're trying to do to be honest
[15:43:31] <Nodex> Lipathor : please pastebin your piece of code , I can only assume your code is deleting the object
[15:45:37] <liquid-silence> Nodex, so if I flag book[0] as deleted I do not want it to be in the result, when I select the product, makes sense?
[15:46:06] <Nodex> Lipathor : without seeing your code I cannot help you sorry so again good luck to you ;)
[15:48:56] <Nodex> this is the only chance you have at mongo returning it without the array elements http://docs.mongodb.org/manual/reference/operator/projection/positional/
[16:00:28] <tavooca> I mongodb during m101, but doing the task HW3 python gives me 500 error blog.py I'll be doing wrong?
[16:48:30] <shantanoo> hi, is it possible to limit the size of the db/collection? (in standalone and shard configuration)
[16:49:01] <shantanoo> was going through http://stackoverflow.com/questions/5054628/limit-mongodb-database-size
[16:57:32] <azathoth99> I am looking for that answer? I mean disk space wise are u supposed to jsut give mongo lots and hope?
[17:01:13] <tomasso> is there some way to access mongo db from an android lib? as if it were a web service..
[17:30:22] <mongonewbie2387> Hi guys, is there a way to do something like what I'm trying to do here: http://pastie.org/8463181 ? Basically there is an entry in my collection with the field "astring" and it has a value "asdasd(blahasdasdf" and i want to replace the "(" with " ( "
[17:30:53] <mongonewbie2387> the only way ive seen to do this is to iterate over each document, get the value out into a variable and change it, then insert the variable back into the entry
[17:40:10] <tomasso> i have a collection with lat and lng values.. im reading on geospatial indexes and found out they use GeoJSON objects, that are like { type : "Point" , coordinates : [ <longitude> , <latitude> ] } . Is there some way to specify my index to use the existing lat and lng keys, and not the GeoJSON point? or I need to convert the collection from lat and lng to GeoJSON point respectively?
[17:58:49] <jpiche> I was hoping to be able to run mongodb on a rasberri pi, but arm isn't supported (see SERVER-1811). It looks like the Scons file for V8 is a big part of what is limiting it, but vanilla V8 compiles just fine on arm. Anyone have any insights?
[18:00:03] <Derick> jpiche: mongodb doesn't run on ARM because we don't support big endian platforms yet
[18:07:30] <jpiche> Derick: really? is there a lot of little endian specific code? When I was looking through the code, it looked to me like the asm blocks might have been the big holdup
[19:54:05] <ph88_> in SQL i have a long table and i need to do WHERE Color = green and then get the unique values from the other columns. Would MongoDB be fast at doing this or is it better i stay with SQL ? (it's ok if i need to change data structure)
[23:14:35] <foobar2k> eg, my chunk's bounds are :bounds=>[{"project_id"=>-1696054874687179387}, {"project_id"=>-1672730701517160081}]