Log file Viewer
#mongodb logs for Friday the 1st of June, 2012
[18:29:35] <jaraco> I've added pmxbot to the channel. Logs at http://chat-logs.dcpython.org . pmxbot is also set up to relay twitter messages to the channel. If that turns out to be undesirable or too noisy, don't hesitate to ping me and I'll turn it off.
[18:39:57] <wereHamster> jaraco: then you need to mention the bot in the topic, otherwise it's agaist the freenode TOS
[18:41:29] <jaraco> It may be that it notifies when one connects to the server. No, it's got to be the channel.
[18:46:20] <jaraco> That's the first news - subsequent entries will only happen when there are new tweets.
[19:05:07] <GRiD> question for core devs: are the error codes in core server/tools assigned in any particular way? i see they are tracked in docs/errors.md
[19:15:42] <GRiD> kchodorow, ok so if i were to add a new one, pick the next highest? fyi i'm looking at submitting a patch for SERVER-5912
[19:18:14] <kchodorow> GRiD: if you set it to 0, scons will automatically insert the next # on compile
[20:27:45] <infinitiguy> if I require auth on my mongoDB but I'm creating a new database - how do I create the initial user if auth is required?
[20:59:17] <westoque> So I have a field wherein I store immutable JSON data, I was thinking of storing it as flat .json files, it would make it faster for fetching in terms of the http server in contrast to mongodb.. Would you guys still recommend mongodb for my use case? Whats the advantages if I do? Thanks
[21:00:16] <westoque> So I have a field wherein I store immutable JSON data, I was thinking of storing it as flat .json files, it would make it faster for fetching in terms of the http server in contrast to mongodb.. Would you guys still recommend mongodb for my use case? Whats the advantages if I do? Thanks
[21:18:27] <SkramX> i have a map reduce that goes through all objects and extracts the event's location and counts all of them and puts the _id and count (as value) into a new collection. Would I use a finalizer to drop all records where value is less than 5?
[22:01:00] <SkramX> i have a map reduce that goes through all objects and extracts the event's location and counts all of them and puts the _id and count (as value) into a new collection. Would I use a finalizer to drop all records where value is less than 5?
[22:02:24] <zirpubolci> i haven't used mongodb MR yet. but i think that might be right. or just make the reduce use a query for > 5.
[22:03:53] <SkramX> and if all the emits are done on one server.. i think the emits are what take so long so modifying the reduce doesnt really help
[23:05:14] <tystr> how would I query and return only a single document in an array of embedded documents?
[23:06:55] <dstorrs> do you want a random document, just to verify something is there? do you want to match a set of criteria? if the latter, what? etc.
[23:07:44] <tystr> db.collection.find({ "embedded_docs._id": ObjectId("4fc84597ab3c453633000002") });
[23:08:36] <tystr> rather, db.collection.find({ "embedded_docs._id": ObjectId("4fc84597ab3c453633000002") }, {embedded_docs:1);
[23:09:36] <dstorrs> have you read these? http://www.mongodb.org/display/DOCS/Dot+Notation+(Reaching+into+Objects) and http://www.mongodb.org/display/DOCS/Advanced+Queries#AdvancedQueries-ValueinanEmbeddedObject
[23:10:33] <tystr> and we're porting our app from mysql to mongodb, so I'm still trying to wrap my brain around the whole scheless design thinking
[23:11:36] <tystr> I've embedded these docs b/c usually when showing content to the user I want to pull everything
[23:12:12] <tystr> but when it comes to adding/editing single embedded documents (in the backend, for example) I'm not really sure how to do it, or if I'm going about it the proper way
[23:13:48] <dstorrs> this is where my "what are you trying to achieve" question comes into play. What are these embeddeed docs and why embed them?
[23:14:07] <dstorrs> The more I talk to people on channel, the more it seems that embedding is often a bad design choice
[23:14:42] <dstorrs> { "_id" : ObjectId("4fc94bc2586d3356a8e56e4c"), "vt" : "not_real", "foo" : { "bar" : 1 } }
[23:16:43] <tystr> db.collection.find({ "embedded_docs._id": ObjectId("4fc84597ab3c453633000002") });
[23:19:19] <dstorrs> the danger with embedded docs is that they are harder to update, and that unless they have a clearly defined limit, you will eventually blow the 16M doc size
[23:20:48] <dstorrs> embedded docs are good for things like "here are all the line items that are part of this invoice"
[23:20:49] <tystr> this is basically a cms type of thing, so I know the document size will not be an issue
[23:47:54] <callen> should I just add an objectid field to them? if so, how do I enforce uniqueness across the list of embedded documents?
[23:49:37] <callen> { "top_level_doc": blahblah, "list_of_things": [{"innerdoc":data, "_id_field_here?":id}, ...]}
[23:50:00] <callen> the unique ids for the embedded documents only have to be unique to that top level document
[23:50:58] <callen> it's in production, but I realized I couldn't keep keying across multiple fields of data in the list of embedded documents to identify them
[23:52:17] <callen> it used to be that they could be one of three types of meals, but that's no longer the case, arbitrary meals are expected now
[23:52:18] <dstorrs> I would probably use epoch because it makes range searches easy and fast. but that's me.
[23:52:47] <callen> what's the "usual" way to uniquely identify an embedded document and enforcement uniqueness within the root document?
[23:54:46] <dstorrs> db.coll.save({ user : "bob", meals : { 2012-jan-18 : { ... }, 2012-jan-19 : { ... } } })
[23:56:25] <callen> the meals don't have something like a date to uniquely identify them, because they're already---yes of course I do.
[23:56:51] <callen> How do you uniquely identify, using something unrelated to the data itself, an embedded document?
[23:57:39] <dstorrs> first, lose the 'tude. I'm trying to help you, and to do that I need to understand the problem. Often questions like this relate to "the problem goes away if you change the data model", which is why I'm asking.
[23:57:58] <callen> and I can't change the data model because we can no longer assume the user has a unique type of meal
[23:58:33] <callen> if this was MySQL, I would've known the answer to this ages ago..."just use a guid or pk"
[23:59:30] <callen> arbitrary list of embedded meal documents, need to uniquely identify them, nothing innate to the data model to do so...leaving me...needing to add something to uniquely identify them
[23:59:46] <callen> or generating my own guids and hoping for no collisions / checking for collisions.