Log file Viewer
#mongodb logs for Monday the 6th of July, 2015
[00:43:43] <a|3x> i heave a bunch of threads running, each with their own ScopedDbConnection, but they still manage to step on each other (query returning results when there should be none)
[02:25:59] <atnakus> Is there some kind of pager function in mongo shell which allows you to scroll through the results both up and down?
[04:00:08] <Jonno_FTW> how do I make it so mongod only allows collections from authorized users? ie. you have to have a username/pw to access the db
[04:05:30] <Jonno_FTW> haha It was allowing anonymous local connections, but now I checked a remote client and it won't let me in
[07:41:06] <xDHILEx> I just posted this on SO if anyone can lend a hand. http://stackoverflow.com/questions/31239617/twilio-account-verification-example-app-typeerror-on-server-restart
[09:45:33] <pamp> Im missing something in my connection string ? client = new MongoClient(@"mongodb://user:pwd@server:27018/admin")
[10:09:00] <CharlesFinley> Hi, why does this query return 0 results? db.getCollection('companies').find({_id: '559a4c7c18b5df6b032b353f'})
[10:11:36] <d-snp> CharlesFinley: does db['companies'.find({...}) work? of db.companies.find({..}) ?
[10:12:32] <CharlesFinley> here's the fix: db.getCollection('companies').find({'_id': ObjectId('559a4c7c18b5df6b032b353f')})
[12:46:25] <pschichtel> Hi! I'm interested in implementing a paginated list with a mongodb. the collection will have a lot of elements, so I thought about implementing this using seek-pagination. In MySQL I'd use an auto incrementing index field for this, what could I use with mongodb?
[12:59:44] <StephenLynx> and afaik, documents are already sorted by default by their insertion order.
[13:08:29] <deathanchor> sorting on an _id field that stores ObjectId values is roughly equivalent to sorting by creation time.
[13:23:39] <pschichtel> I also prefer strongly typed languages. I like my shit to crash when I do stupid things
[13:28:45] <EXetoC> and it's nice not to have to worry about key lengths and so on. I haven't had to care yet though
[13:32:33] <deathanchor> at least there are things like mongoengine where you can add application side enforcement of scheam
[13:48:40] <EXetoC> modern statically typed languages prove that static typing doesn't imply less productivity
[14:09:01] <Nomikos> in the CLI client, if you've done a db.foo.find(), it returns a DBQuery object - is there some way to run that again?
[14:49:21] <jubjub> If i do a query on {name: "john", age: 17}, does mongo perform the query on all docs and short circuit is the name isn't john? Do they check the query params in order?
[15:10:31] <Steve-W> Hi all. I'm using ruby to list collections in a Mongo 2.4 database, and it is only returning the first 101 items. Is there some way of enumerating them all?
[15:18:29] <jubjub> If i am querying to see if a point lies within a polygon which query is fast? $geoWithin or $geointersects?
[15:47:34] <na8flush> hey all, i'm not experienced enough with mongodb to know if this is a bad design pattern or not, so i'm hoping someone might be able to give me some feedback
[15:49:12] <na8flush> on our app, we store user details in an object called UserInfo. When that user authenticates, we generate a token for them, and that gets added as a property to the record. but here's the part i think might have negative performance - we save that object as an entirely new collection, UserInfoWithToken, duplicating all of the info from the UserInfo record
[16:34:08] <asturel> hi, after a while my mongodb server keep throwing 'DBClientCursor::init call() failed'
[16:35:08] <asturel> oh may i found why.. Mon Jul 6 18:31:32.809 [initandlisten] connection refused because too many open connections: 819
[16:36:23] <StephenLynx> so you don't have to worry about concurrency, you just reuse it until the application stops
[16:36:44] <symbol> Can someone give me an idea of what sort of issues I could run into with batch updates not being atomic? Does it have to do with interleaving?
[16:37:33] <StephenLynx> IMO, it could perform just some of the updates and then you wouldn't know which ones ran and which didn't?
[16:40:00] <StephenLynx> I have a module that holds the connection and open references to collections on boot.
[16:40:17] <StephenLynx> it ensures all indexes, gather all pointers to collections and then just serves these pointers.
[17:59:37] <ToMiles> Any reason a db.repairDatabase of a 10GB db would be allocationg and filling with zeroes more then 100 ( and counting) data files in the _tmp folder diuring this process?
[18:24:04] <danijoo> i deleted a ~ 70gb collection from a database. the server is pretty low on disk (~80% in use, all by mongo). I know that i cant reclain that 70gb without doing a database repair (what i cant do because not enough space on disk).
[18:24:45] <danijoo> will the other collections from the same db start to use that 70gb or are they really "dead space" now
[18:31:04] <ToMiles> just killed before it filled up my whole disk with useless zero fille data files
[18:53:41] <MacWinner> hi, i currently have a mongo replica set on a 3-node dedicated c;uister.. i'm thinking that I may just want to outsource this and use a mongolab type of service. I don't mind paying a premium for it, but don't want to get gouged either. Any recommendations on this? I'm also open to maybe a nice management tool that lets me deploy and scale out the cluster onto aws or google cloud
[18:54:22] <StephenLynx> coderman, I have a hunch that a single query will probably be ran on a single thread
[18:59:21] <akoustik> cheeser, MacWinner: might be the other way around. http://stackoverflow.com/questions/8331099/what-is-the-javascript-engine-that-runs-mongodb-shell *shrug*
[19:00:45] <MacWinner> akoustik, looks like v8 became default in 2.4.. http://docs.mongodb.org/manual/core/server-side-javascript/
[19:10:07] <akoustik> anyone familiar with running a replica set with SSL? i think what i want is 1) all communication between replica set members should be via SSL, and 2) client connections should be non-SSL. looks like all i need is to run with net.ssl.mode == "preferSSL" and provide my key files. am i missing anything obvious?
[19:38:18] <amitprakash> Hi, I have a collection with documents of the format {'loc': <string>, 'date': <ISODate>, 'in': <boolean>}
[19:39:05] <amitprakash> I want to calculate (sum(date) where in = 0) - (sum(date) where in = 1) grouped by loc
[19:53:51] <amitprakash> so I can use $add in mongo aggregation with date and a number.. any way to add two dates?
[20:00:30] <iszak> I am running some queries on mongodb in which it has roughly 90k documents / day and this seems to very slow even for a few days of data with a range (date) is mongo no good for querying?
[20:04:29] <akoustik> well, you'd want to be indexing on the date field, so one thing would be to make sure that you are doing so.
[20:04:57] <iszak> yeah I will check that but in general everything I have heard is mongo is rubbish for querying
[20:08:39] <akoustik> iszak: regarding upgrading - getting to 3.0 gains you access to pluggable storage engines. i've not messed with them, but it looks like WiredTiger for example might emphasize high performance queries.
[20:09:32] <iszak> joannac: I can't do it atm, it was at work, but I will remember this for tomorrow.
[20:09:44] <joannac> from what you said, it sounds like you don't have the right indexes. Upgrading to 2.6 or 3.0 is not going to fix missing indexes
[20:11:47] <joannac> at which point, your disk setup is probably more important than the database you try to use
[20:11:49] <EXetoC> akoustik: does that mean you can use hashing rather than comparing key contents for example?
[20:12:37] <akoustik> EXetoC: no clue, tbh. don't know how things really work under the hood with any storage engine.
[20:14:19] <amitprakash> how do I update a field in all documents depending on values from another field in the same document?
[20:20:57] <joannac> do you actually have a connection to a database? Can you find the same document using the mongo shell?
[20:23:39] <joannac> I would like to see where you connect to the database, and also the same query in the mongo shell, and also a find() with no arguments in python
[20:24:07] <joannac> to prove that you do have a connection, the document exists, and you're connecting to the right server/db/collection in python
[20:25:10] <Havalx> akoustik, akoustik, cheeser: expression d does not return the expected result. {"x": 7} expressed as string. find_one({"x": 7} does return the document, but not find_one(d). A connection is active and the document has already been created.
[20:26:53] <akoustik> Havalx: well... line 8 is looking for a property "d" of "db" - that property name doesn't have anything to do with the string you've assigned to d.
[20:29:32] <Havalx> akoustik: Thanks for nothing such, I've concerned with that. What is the proper way of forming a connection? I would not expect there to be a python problem as variable expansion didn't see off.
[20:31:40] <akoustik> Havalx: sorry buddy, haven't used the python driver. hopefully one of these folks can help. i gotta go - peace, all
[20:31:46] <Havalx> akoustik: I read a few chapters of documentation and online comments, and formed a client instance
[20:36:11] <topwobble> How do you set the journal path in the mongo config file? I don't see an option for that (using the non-YAML config)
[20:37:54] <joannac> topwobble: ? don't think there's ever been a way to set a different path for the journal
[21:14:05] <ToMiles> to follow u on earlier message about ever expanding set of preallocating data files (>20x orginal db size) during database repair -> mongod --repair or db.DatabaseRepair() same problem or is that to be expected?
[21:34:24] <ToMiles> daidoji: I knew it need soem space littel more then double, but my db consists of 7 datafiles taking up about 10GB total, but the _tmp folder has >100 fiel sof 2GB each and it just keeps to endlessly increasing ( prob untill I'll have ot kill it again to avoid full disk)
[21:42:14] <ToMiles> the pastebin is what is running now, earlier I had to cancel the repair when it was filling up the disk, it had files up to ceilometer.133 and more in /var/lib/mongodb/_tmp_repairDatabase_0
[21:43:03] <ToMiles> if I cant get it workign I'll prob just clear the collection and deal with a 2 weeks loss of metrics for my OpenStack cloud
[21:49:06] <joannac> also, maybe increase log verbosity for like, 2 mins? just to see what's actually happeneing
[21:49:31] <joannac> definitely weird though. are you reclaiming space or is there corruption in the original?
[21:51:12] <ToMiles> I suspected possible corruption since we had a RAID card failure, that has now been replaced, I wanted to make sure
[22:07:21] <ToMiles> just that repeating for more and more data files cant see anythign out of the ordinary
[22:12:51] <ToMiles> thanks for your assistance, good thing is its not super important data, openstack metrics data, we can afford to loose it if no alternative, but still would be nice to figure out at least a general idea has been goign wronf
[22:19:49] <joannac> 'kay. my prediction is next steps would involve looking at the files and data structures directly
[22:20:44] <ToMiles> just run mongo as Ceilometer DB backend as a part of the OpenStack toolset of our private cloud to collect metrics so nto sure at this juncture it worth further time and effort if a start from clean is an option
[22:25:52] <ToMiles> thanks for you help tho and I'll remember the option of getting that support contract if further down the road the data collected in MongoDB becomes a critical part of our production setup
[23:13:20] <joelkelly> Hey guys ... question about populate. If the DB was built without the populate ref ... can I still provide a ref for the schema?