Log file Viewer
#mongodb logs for Tuesday the 15th of September, 2015
[01:11:06] <silasbarta> so is the mongo v3 logger supposed to write the full document that gets created?
[02:06:28] <tejasmanohar> hey guys im modeling out a referrals system where i also want to be able to rank users based on their referrals
[02:06:54] <tejasmanohar> should i be using a counter cache for this? i feel like the other method would be quite inefficient.
[03:10:31] <JaVaSan> Hi, is there any way to create a query to retrieve a nested array as the root document in response? (i.e. I would like instead of {"fieldArray" : [...] } something like {[...]} here fieldArray is the nested field which is an array of documents)
[03:18:19] <JaVaSan> so, instead of {"fieldArray" : [{...}, {...}] } I would like to retrieve [{...}, {...}].
[09:26:08] <carver404> hi, i'm getting a 404 while installing mongodb 3.0.x on fedora.. any headers?
[09:26:38] <carver404> the baseurl https://repo.mongodb.org/yum/redhat/$releasever/mongodb-org/3.0/x86_64/ gives 404
[11:16:10] <Siyfion> I keep getting a "MongoError: cursor killed or timed out" when performing a simple .find({}) on a collection with around 10MB of data in it... Any idea why that might be?
[11:31:49] <tibyke> im trying to find in a db and order the results by the multiplication (or whatever mathematical relation) of two fields. should it be whatever.find(..., ...).aggregate() or whatever.aggregate().find(..., ...)? neither of these works actually :)
[11:43:04] <carver404> Siyfion: http://docs.mongodb.org/v3.0/reference/method/cursor.addOption/#DBQuery.Option.noTimeout
[11:43:44] <Siyfion> I think it's that I'm doing too much stuff concurrently with the same db client
[15:37:52] <Owner> can / does oplog potentially leak data between logical databases if a spy had access to the 'oplog' data
[15:39:53] <cheeser> but yes, the oplog might have sensitive information in it so you should restrict roles accordingly.
[15:40:32] <Owner> so any 'application' that uses the oplog, the owner of that application could potentially see other applications data
[15:45:30] <zacharypch> HI - is there a configuration option in mongo3 for hyphenation as a delimiter in text indexes? I want to keep english as the language choice, just remove the treatment of hyphenation as splitting since my dataset has a lot of hyphenated terms
[15:49:46] <Owner> cheeser, well for performance reasons this app needs the oplog, but for economic reasons, they only want one mongo cluster for all the environments to share (with seperate dbs)
[16:14:29] <deathanchor> what's the right way to have two member replset with no 3rd member for the short term?
[16:24:29] <mensgis> hey all! I'm switching from a application that created id's like kScTJM9n+g2TUU9eVfQu .. To do things right, right from the beginning, I'd like to use valid ObjectId's. Do you have any idea on how to change existing data easily?
[16:26:06] <mensgis> i am speaking about the _id-attribute.. will removing it just create a new one? if so I might go for updating references manually
[16:27:50] <mak``> Hello :) If I don't care about sequentiality of numeric IDs, but I want uniqueness and for the service to not be a SPOF, how can I do this? I need numeric 32bit integer IDs for MongoDB and sync it with SQL Server in a distributed system.
[16:30:35] <deathanchor> RDBs syncing with NoSQLDBs (or vice versa) is like trying making special sauce: goes bad if left in the sun too long.
[16:36:22] <mak``> StephenLynx: the current system generates the ID in SQL Server and persists it in MongoDB, but it is a nightmare of message queue and we need to change the master from the legacy system to the new system and ultimately switch off SQL Server
[17:05:54] <cheeser> it'd mean an extra db trip but it'd be durable across application runs and unique across a cluster (provided you have more than one instance of the app running)
[17:21:02] <mak``> thanks StephenLynx and cheeser I am going to think about it tomorrow as I am not functioning anymore today :D
[18:37:53] <ChALkeR> But could someone point me to some info on what was going on with all the js engine switches back and forth?
[19:34:33] <silasbarta> So is there any reason why mongo v3 writes entire insert commands to the logs, showing the entire document?
[19:38:17] <silasx> Sorry if I’m repeating myself, but is there any reason why the mongo logs would display an entire document that got inserted?
[19:39:11] <silasx> ahhh … thannks deathanchor. Is that documented anywhere? And what’s the setting that governs that?
[19:39:59] <deathanchor> http://docs.mongodb.org/manual/reference/configuration-options/#operationProfiling.slowOpThresholdMs
[19:43:10] <silasx> Thanks a ton guys! My setting is currently {was: 0, slowms: 100} … so that means the profiling level is 0, but it will still log 100ms ops?
[19:45:32] <deathanchor> silasx: 0 only writes to logs, 1 writes slow ops to profile collection (capped collection), 2 writes all ops of that db
[19:53:18] <StephenLynx> hey, findOne can perform faster because it will stop when finds a document, right?
[19:54:32] <silasx> @deathanchor: Ah, okay, the 100ms threshold holds all the time. Yeah some of these may be because of taking a long time, some are big inserts, but still that seems like a long time. Will have to look into it but this explains a lot.
[19:55:18] <StephenLynx> in the meanwhile, if I want to check if TWO documents exist, a count will be faster than a find and toArray because it will perform less work at each document found?
[19:55:19] <deathanchor> silasx: I was curious how big these docs were and how slow they were. I deal wiht some big docs also
[20:09:21] <silasx> @deathanchor if your’e still around, a bunch of these are insertions of docs around 3000 bytes in length … doesn’t seem like they’d take a long time
[20:12:50] <silasx> @deathanchor: most of those are 1-2, but there’s also this: timeAcquiringMicros: { w: 1331327 } }
[20:13:56] <deathanchor> yeah I still don't understand how if the write lock wait was 1.3 seconds how the op time is less than that (0.8 seconds)
[20:15:36] <silasx> @deathanchor well, a few: 1) We don’t like data being written to logs for security reasons, and 2) now that we see these are long ops, why are they taking so long
[20:19:50] <silasx> wait what about slow ms? What would I set to a large number? Looking up contention
[20:21:27] <beekin> When using the aggregation pipeline, I read somewhere that's it best practice to $match and $project early on in order to trim data. Does this mean, that I should always be projecting only what's needed for the entire aggregation?
[20:25:52] <deathanchor> beekin: I have hit this error: 16MB doc size limit for results, limit it to a smaller number
[20:26:08] <deathanchor> or even worse: exception: terminating request: request heap use exceeded 10% of physical RAM
[21:15:18] <diegoaguilar> Hello, is there a way to drop a database and not to lose the created indexes
[21:16:26] <Owner> jeez im trying to run mongodb as the mongodb user from root, but its not doing anything or logging anything :(
[21:17:15] <cheeser> i'm sure there's a question in there somewhere. some details of what you tried would be helpful.