Log file Viewer
#mongodb logs for Thursday the 22nd of January, 2015
[00:42:32] <adoming> hey guys can you find a sub sub document by objectId in a db.collection.find()?
[00:43:13] <cheeser> subdocuments don't typically have IDs but yes you can query by properties of subdocs.
[00:44:20] <adoming> in other words i have a grandParentDocument ref parentDocument ref childDocument
[00:45:02] <adoming> i want to in one query db.grandParentDocument.find() by a property of childDocument
[00:58:30] <adoming> OK so to reiterate, I am doing a query on this schema http://pastebin.com/TnLm4syg and I am referencing other schemas which in turn are referencing another schema by objectId. My question is how I can db.colleciton.find() data from a referenced, referenced schema (sorry I'm a noob if this sounds dumb).
[01:22:24] <morenoh149> what does findOne return if no document matches? http://docs.mongodb.org/manual/reference/method/db.collection.findOne/
[01:40:03] <morenoh149> dammit it differs http://mongodb.github.io/node-mongodb-native/2.0/api-docs/
[03:02:04] <morenoh149> 3 operations give a write lock http://docs.mongodb.org/manual/faq/concurrency/#which-operations-lock-the-database
[05:28:58] <Climax777> hi all. philosophical question: I get what mongodb is good for and not good for. but what can one say about mysql? what is it good for and not good for?
[05:39:21] <arussel> I have document {a: 1, b: 2, c:3}, if I add an index on a,b will it be used of query on abc ?
[06:07:57] <DragonPunch> Message.findOne({Email: asdf@gmail.com}, { sort: {'created_at' :-1} }, function(err,data) { // Do stuff });
[07:18:22] <x44x45x41x4E> Hi, I've got a question. I'm backing up a MongoDB instance that is on production using mongodump, will my dump be more reliable if use oplog option? Thanks. http://docs.mongodb.org/manual/tutorial/backup-with-mongodump/#point-in-time-operation-using-oplogs
[07:22:19] <Viesti> hum, thinking of representing a tree structure in mongo as just one document, having operations to add and delete nodes
[07:23:24] <Viesti> so thinking that maybe child nodes nodes should live under ObjectId keys, instead of in an array
[07:23:29] <x44x45x41x4E> joannac: Nope, it's on a VPS. Just to be clear, what do you exactly mean by 'standalone'?
[07:24:27] <joannac> is it a replica set, sharded cluster, or a single mongod instance (e.g. standalone)
[07:31:00] <x44x45x41x4E> joannac: Alright, thanks. I just thought there are benefits to using the oplog option. Newb to MongoDB. Thanks. :)
[07:31:50] <joannac> there are, but since you're not running a replica set, you can't take advantage of them... :p;
[08:00:10] <morenoh149> finally the answer to all my questions http://mongodb.github.io/node-mongodb-native/2.0/tutorials/crud_operations/#toc_5
[08:55:03] <gonace> Hi, is there a way to verify if one have used: db.runCommand( { logRotate : 1 } ) apart from checking the log dir?
[09:00:16] <gonace> Hi, is there a way to verify if one have used: db.runCommand( { logRotate : 1 } ) apart from checking the log dir?
[09:03:19] <Yogurt> Yeah. (: Iam here too. I mean for help. And i asked so many people. So, noone answer. Btw iam hanging around for chat. Coz, it's been over 20 years to chat with mIRC in my life.
[09:53:13] <Tomasso> hello, I read documents from a collection, I processed and modify them, I also remove the _id field, and insert it into another collection, I get DUPLICATE KEY. I do the same, but insert in a collection from a different database, and I get the same dupplicate key... I'm lost on this. I read there was a bug on smth like that, I use mongo 2.6.1
[09:54:17] <Tomasso> also i don't understand since I removed _id field.. I print the doc to stdout and _id is not there.
[10:57:53] <Yogurt> aaearon sorry bro. but oneone is here for help. you have to believe me. i tried so much (:
[11:34:54] <temhaa> Also I have another question. in mongo server there are some problem. mongo slave output: http://picpaste.com/mongoslave-7fl71XCc.png
[12:11:52] <temhaa> joannac: Actually our problem is master log very big but slave logs very small. I need check master slave repl. is true.
[12:53:08] <Tomasso> after modifying an object and inserting it to another collection, i get Duplicate key. I tried to delete the _id entry, and same error... I tried ticket["_id"] = BSON::ObjectId.new and again same error.... i dont know what to do
[12:58:51] <Tomasso> im generating the _id by code, this is what it prints now {"_id"=>BSON::ObjectId('54be4d05c783c41d5c000001') ..........
[13:02:54] <StephenLynx> I think you are messing up something and not actually removing the id. theres no way it would says it has a duplicate field when the object does not even contains such fiel
[13:07:24] <Tomasso> yeah.. you were right xD I catch the exception and was printing duplicate key... not the real error message xD
[14:15:05] <Tomasso> how do i get the size of a document that is in the bson format? if possible from ruby mongo driver better..
[14:27:23] <StephenLynx> and if you can get the size of an object, you could use the same for these bson objects.
[15:20:28] <StephenLynx> when you think about what would take for someone to place a MIM attack on that
[15:21:17] <StephenLynx> of course, but what would take for someone targeting you to place the attack?
[15:22:15] <StephenLynx> ok, now what would be the costs for them to target every single person and not get caught so everyone would become aware of it?
[16:35:39] <the_drow> Can we drop a collection to free disk space without repairing which requires downtime?
[17:23:57] <neo_44> cheeser: one of first things I look at in any architecture is fault tolerance.....with single node you have to fail over, so you have a single point of failure
[17:24:57] <neo_44> and from client side a single node is the same as a replica set...so there are no code changes....just connection string needs to be updated to have seed list
[17:57:22] <jayjo> I'm using mongoengine to interact with mongo, is it possible to return an object in stead of a queryset when I query the database?
[18:22:07] <imachuchu> so I need to implement a changelog functionality to an existing collection, tracking only modifications to individual documents. I'm thinking of doing it by appending a subarray on each document that stores a subdocument for each change (so who, when, what, old value, and new value). Since I'm a bit new to MongoDB I'm wondering does this sound good or is there a brighter way?
[18:25:21] <neo_44> jayjo: Client.objects.get(__raw__={"_id": oid}) this will return single document
[18:27:59] <neo_44> imachuchu: could do you need to know what changes or just that something changed? ... is this to roll back a change?
[18:34:13] <GothAlice> imachuchu: If one needs to track document changes over time, one could watch the oplog the same way MongoDB replication does, making note of updates across the desired collection. Then, to track additional information (beyond the literal query used to make the change) you can use $comment within the query to pass additional data through. I do this to pass the User ID who requested that update.
[18:35:26] <GothAlice> https://github.com/cayasso/mongo-oplog is an example library to provide this type of low-level oplog functionality.
[18:36:38] <neo_44> Depending on the use case I agree with GothAlice....this would be great for auditing if needed
[18:37:08] <imachuchu> neo_44: it's so that a manager can audit to see what all has been changed in a record and by what employee (essentually)
[18:37:36] <imachuchu> GothAlice: that's almost exactly why I figured I should ask here, let me take a look and see if that's closer to what I want/need
[18:39:22] <neo_44> imachuchu: I would add the $comment as GothAlice suggested on all updates to the record
[18:40:03] <neo_44> imachuchu: if you use a service layer it could get tricky because the WHO would be the service layer not the person...FYI
[18:42:45] <neo_44> imachuchu: you could also use a seperate capped collection to write the old document and who changed it....this would clean it self up over time but would give you insight for a certain time period
[18:47:46] <imachuchu> neo_44: that was coming up on my next question, is there any reason this shouldn't be stored inside a seperate collecion inside MongoDB? While there shouldn't be too many changes (>100, most likely >50) per record I'm still a firm believer in preparing for the worst
[18:48:52] <saml> {"x": {}} how can I query for docs where x has at least one key? such as {"x":{"y":1}}
[18:49:07] <imachuchu> the "who" will be a bit more difficult, since it's who in the application changed the record, not who in MongoDB, but I'll look into implementing mongo-oplog and see what I can find
[18:49:34] <neo_44> imachuchu: are the changes atomtic and replace the entire document? or a lot of little updates?
[18:50:24] <neo_44> imachuchu: I would use seperate collection and bucket all the changes so 1 query would return all changes for that doc over time
[18:50:31] <imachuchu> neo_44: since I'm writing the frontend that's doing the changes right now, whatever's best. I believe they are little changes but I can make them all one (I'm using Mongoengine and haven't really checked)
[18:51:31] <neo_44> if you use mongoengine and just save the object (object.save()) it will replace the entire document....so that is multiple chagnes at once....then you could just overwrite the .save() to copy the document to the new collection first
[18:52:09] <neo_44> I use mongoengine but have wrapped it in a data access layer to abstract away the mongoengine piece for the UI/API guys that use my DAL.
[18:54:49] <imachuchu> neo_44: hmm... I see. I'll need to look into what I can get because storing a full copy on every update to the capped collection seems wasteful
[18:56:07] <saml> db.asdf.aggregate({$match:{x:{$ne:{}}}}, .... want to get count for each key x.$key
[18:57:40] <imachuchu> neo_44: duplicating the whole document on every commit when I'm only looking at storing the deltas
[18:58:27] <neo_44> imachuchu: with a capped collection there is no fragmentation and how big are the docs really? It would be easier to just store a copy then try to only store the deltas... .
[19:00:46] <imachuchu> neo_44: true, and the documents aren't that big, so it's not too big of a deal. I'll go implement something and come back here with any questions. Thank you for all of your help
[22:01:27] <FunnyLookinHat> I have a collection of "particles" - and each particle belongs to a beam ( one beam can have many particles ). https://github.com/funnylookinhat/tetryon/blob/master/spec/particles.txt
[22:01:44] <FunnyLookinHat> I chose to embed the beams inside the particles because it will make generating reports and whatnot far faster
[22:02:03] <FunnyLookinHat> But I'm seeing an issue - I want the beam attached to each particle to always be the same -
[22:02:43] <FunnyLookinHat> The only way I can think to do that is, when I create a new particle, read another with the same beam.id and then use that beam information in creating the new particle
[22:13:04] <kexmex> so like, due to non existance of a signed mongodb rpm for centOS, i am trying to build my own. How do I build rpm, after building from source?
[22:21:53] <joannac> kexmex: what? http://docs.mongodb.org/master/tutorial/install-mongodb-on-red-hat-centos-or-fedora-linux/
[22:30:18] <joannac> kexmex: the repo has an rpm subdirectory with .spec files, that should be enough
[22:32:18] <kmaru_p00ps> is there a way to make rs.status() print all status in a line instead of the default pretty() way?
[22:34:34] <kmaru_p00ps> joannac: get all info for hosts in one line, like I can do with the other db commands and parse it in a shell script
[22:35:32] <kmaru_p00ps> parsing pretty() with just basic shell script? and not using other tools..can be done..but not easy.
[22:38:08] <kmaru_p00ps> sorry...don't want to have to use other tools. why is rs.status() so special that it's default is pretty()? *sigh*