Log file Viewer
#mongodb logs for Friday the 15th of January, 2016
[00:47:46] <jbrhbr> nothing fancy really, just slow. not sure if i'm missing something obvious but i thought this would be faster than 600ms
[00:49:37] <Boomtime> that's the entire collection presumably - it looped over every key in the index
[00:52:17] <Boomtime> since you already have this data set.. can you try adding 2 new documents, with values of 0 (zero) and 2
[04:17:41] <Voyage> Boomtime, so mongo has CAP fully? consistency, availability , partition tolerant? You only can be good at 2.
[15:59:50] <metame> btw here's some (academic but short) reading on CAP: https://www.comp.nus.edu.sg/~gilbert/pubs/BrewersConjecture-SigAct.pdf
[16:01:43] <ThunderMuscle> Hi! Small question about c++ client library (legacy 1.1). I did a pretty small test with it and encounter a weird problem. I just insert a simple object with my program. It executes without any errors. when executing db.stats() in the db, it also tells me "objects": 1 (and consecutive execution of my program increments this) but doing db.test.test1.count() or db.test.test1.find() returns 0 resp. nothing. Do I have to set some
[16:01:43] <ThunderMuscle> specific options in my lib that it also can be seen by the 'normal' client?
[16:15:56] <ThunderMuscle> Well - think so :) So This is the code: http://pastebin.com/eRw30ykz and this is my querying on the 'mongo' command line tool http://pastebin.com/LaRT42mf
[17:52:39] <dooderino> Sure, though there might not be an adequate solution, just take a look here http://stackoverflow.com/questions/34811299/mongo-find-limit-each-match
[17:57:36] <dooderino> yes? But it's not a very good solution performance wise if you haven't noticed, I'm wondering if there are any better ones
[18:05:50] <johnjohn101> well i have a collection. does it always return the 1st item that in inserted into the collection?
[18:05:59] <cheeser> you could always pre-aggregate and use $out to store the results in a collection then query that.
[18:20:36] <acedude> I have a pretty specific question: when this query "db.getCollection('checkins').find({"created_at": {"$gte": new ISODate("2013-10-01T00:00:00.000Z")}})" stopped being correct?
[18:21:03] <acedude> it works just fine on Mongo 2.4, but on 3.2 it returns nothing (until I add .toJSON() to the ISODate)
[18:34:23] <acedude> looks like my issue might be driver-specific, but I'm not sure... I'm so desperate I'm downgrating to mongo 3.0, then 2.6
[19:08:51] <acedude> @cheeser like that http://uploads.acedude.pl/2016-01/Robomongo_0.8.5_2016-01-15_20-02-50.png
[19:14:23] <stickperson> i’m trying to use $sum in a projection in mongo 3.2.1, but i’m getting seeing an “unsupported projection” error
[19:32:39] <stickperson> cheeser: the docs say that $sum can be used in projections as of 3.2. i’m using db.myCollection.find(query, projection)
[19:34:52] <stickperson> metame: sure. here it is on the $sum page: https://docs.mongodb.org/manual/reference/operator/aggregation/sum/#grp._S_sum
[19:35:21] <cheeser> that's still an aggregation operator. you're not doing an aggregration. you're doing a find()
[19:35:32] <stickperson> here’s the release notes: https://docs.mongodb.org/manual/release-notes/3.2/#agg-accumulator-availability
[19:38:20] <metame> query projection is pretty light in operators/syntax https://docs.mongodb.org/manual/reference/operator/query/#projection-operators
[19:38:29] <jbrhbr> metame: did you have any more thoughts on my slow count(_query_) issue yesterday? Boomtime said he was testing something but never got back to me
[19:40:55] <metame> if I'm thinking about this correctly since count() is shorthand for find().count()...
[19:43:42] <metame> jbrhbr: don't remember if you have any other props, but also put state_enum: 1 in your projection
[19:56:53] <jbrhbr> yea, i see the changed note at the top. but in the older version, i don't think that syntax supported .count() so i was using .runCommand() :p
[19:57:07] <jbrhbr> and my .count is still 600ms :p c.explain("executionStats").count({ state_enum: 1}, {state_enum: 1, _id: 0})
[19:58:36] <metame> wasn't sure if it'd help since .count is like .find().count() and he's trying to use an index
[20:03:39] <cheeser> i have a collection of just over 9M log records. count() is almost instantaneous.
[20:09:49] <jbrhbr> heh, it's 0ms if i query with state_enum:0 instead of 1 (all of my documents have it set to 1)
[20:10:08] <jbrhbr> it just seems surprising that this index doesn't know how many things it has in each path
[20:11:01] <jbrhbr> like it seems to be doing a full scan of the data just to get the count where this value is 1
[20:11:36] <cheeser> just confirmed that it does indeed have to do an index scan to generate the count
[20:12:36] <jbrhbr> does it have to for architectural reasons or do you just mean it does so currently?
[20:25:41] <saml> "$date" : 1442761653000 "$date" : 1391040000000 "$date" : 1376870400000 so mongoimport chokes on one of these
[20:25:58] <saml> 2016-01-15T05:59:23.782-0500 Failed: error getting extended BSON for document #9: invalid type for $date field
[21:25:39] <saml> The MongoDB JSON parser currently does not support loading ISO-8601 strings representing dates prior to the Unix epoch. When formatting pre-epoch dates and dates past what your system’s time_t type can hold, the following format is used
[21:26:10] <saml> {"date": {"$date": {"$numberLong": "68400000"}}} this works. {"date": {$date: 68400000}} this does not work cause number is pre-epoch
[22:27:31] <digitalvectorz> Ello all. Using mongodb c# 2.2.2 - has the MongoDB.Driver.Builders namespace been relocated or replaced with a more preferred practice?
[22:59:08] <Boomtime> @digitalvectorz: it's still there AFAIK: http://api.mongodb.org/csharp/current/html/N_MongoDB_Driver_Builders.htm