Log file Viewer
#mongodb logs for Wednesday the 24th of June, 2015
[01:26:50] <goldstar> AbuDhar: I can do this with other dbs, why doens't mongo offer the same functionality? Im not going to spend 20min spinning up a box just to test whether a pass works
[08:38:55] <yauh> g'day - I'm inserting geojson data into MongoDB with a 2dsphere index. I noticed that even though I trimmed my coords to have only 4 decimal places the collection stores 17 places
[08:49:35] <yauh> I guess I am looking for a way to configure the bits for 2dsphere, I only found it for 2d index
[10:36:01] <leporello> Hi. I have a collection with createdBy field. How can I get a list of all possible creators from it?
[11:51:04] <circ-user-u0N69> hey guys, is it possible I can specify multiple listening ports on my mongo instance?
[11:55:18] <leporello> iptables -t nat -A PREROUTING -p tcp --dport 10000 -j REDIRECT --to-port 20000
[12:11:24] <StephenLynx> don't think so. I don't think you can't do that for any TCP based connection.
[12:21:58] <leporello> circ-user-u0N69, it's not possible, so you need to redirect all incoming connections on other ports to mongo.
[12:23:51] <deathanchor> leporello: rightly said! firewall or proxy to switch up the port or allow multiple ports
[14:33:32] <d-snp> Hi, I'm having a write throughput problem, every 10-20 writes, an oplog query randomly takes 500-2000ms
[14:34:01] <d-snp> QUERY [conn29] getmore local.oplog.rs query: { ts: { $gte: Timestamp 1435098587000|18 } } cursorid:22303303740 ntoreturn:0 keyUpdates:0 writeConflicts:0 numYields:0 nreturned:1 reslen:293 locks:{ Global: { acquireCount: { r: 6 } }, Database: { acquireCount: { r: 3 } }, oplog: { acquireCount: { r: 3 } } } 1341ms
[14:59:39] <jmeister> does anyone know if I call collMod to change an existing TTL index, does it rebuild the entire index or is it safe to run on a production environment?
[15:10:06] <jmeister> Bump - does anyone know if I call collMod to change an existing TTL index, does it rebuild the entire index or is it safe to run on a production environment?
[16:09:33] <muhammadn> I have trouble benchmarking ruby code (not ruby on rails) to see how fast it can insert data.
[16:10:44] <muhammadn> correct gist is here for mongodb benchmark code in ruby: https://gist.github.com/muhammadn/b16a6b5d8396a3ee8283
[18:06:39] <deathanchor> ooo... is there an easy way to know which shard has a specific key without using sh.status({ verbose :1 })?
[18:08:03] <deathanchor> just trying to find the right shard without scraping through sh.status() which cna be big
[18:08:08] <darius93> I have a question, what are the chances of there being two of the same uuid within a collection? I do slightly rely on the uuid but may just move away from that and into my own id system
[18:14:57] <deathanchor> darius93: there are already several ways to generate uniq ids, don't make your own
[18:17:35] <darius93> lol deathanchor i meant for my internal use. I dont mean generating uuid itself though if i had to make my own
[18:21:56] <deathanchor> darius93: yeah we use a collection just to get a new objectId for our internal ids
[19:24:53] <whaley> darius93: +1 on ObjectId also. It has some other nice properties besides uniqueness - for instance, it contains a timestamp as a property, which completely obviates the need for a "created" field on your document
[19:25:54] <whaley> granted, that's in seconds and not millis... and probably won't work after the year 2038, but it's still a nice little value add
[19:55:12] <deathanchor> cheeser cheated.. it's already all here: https://en.wikipedia.org/wiki/2038
[20:16:59] <devSeb> Hey mongo-folks! =) Learning mongodb since a while, with Node.js. Now I created a function using MongoClient that can drop a DB using db.dropDatabase() (not from shell, from a node.js app). I'm failing at creating a function that creates a new DB though. Any pointers on this? I can't find docs on using copyDatabase() except from the shell.
[20:27:24] <cheeser> well, you have to write to it first. or run "show collections" (the listCollections command at least)
[20:32:25] <devSeb> StephenLynx: Exactly that, is what i mean. I know mongo works like that from the shell - will have a go for it!
[20:33:09] <StephenLynx> working with mongo using the node.js driver is pretty much the same as working with the terminal.
[20:34:35] <devSeb> StephenLynx: I see... But there's no copyDatabase function I can access, then? Like in the shell with db.copyDatabase. I have to read and write manually?
[20:36:18] <StephenLynx> never had any need to use commands on an application that weren't implement.
[20:38:08] <devSeb> No... its just me trying to learn mongodb better. I'm trying to "reset" my dev db before unit tests... But maybe there's some smarter way of doing that?
[20:39:37] <devSeb> Ok, or really the tests doesn't matter, but im trying to do this to have a fresh db on every build...
[20:48:10] <devSeb> StephenLynx: thanks man, really appreciate it! Looking through more of your code too... And gonna read up on indexes now. bbl =)
[20:48:27] <macwinner> will an index on an attribute like an md5 hash have a significantly different performance or memory usage characteristic vs an index on ObjectId?
[21:03:53] <greyTEO> I know this goes against mongo but what is the best way to compare 2 collections?
[21:40:11] <macwinner> ie, if I drop userid_1 will anything that depended on it be optimized with testid_1_userid_1 ? or does the ordering matter?
[22:41:40] <fxmulder> so I've added a new replica set member and I noticed it's created more data files for the database than on the replicas its replicating from, why might that be?
[23:16:30] <d-snp> ah I keep missing the cool people, or they dont think my question is cool enough :(
[23:16:48] <d-snp> anyone know why oplog queries like this could be slow? QUERY [conn29] getmore local.oplog.rs query: { ts: { $gte: Timestamp 1435098587000|18 } } cursorid:22303303740 ntoreturn:0 keyUpdates:0 writeConflicts:0 numYields:0 nreturned:1 reslen:293 locks:{ Global: { acquireCount: { r: 6 } }, Database: { acquireCount: { r: 3 } }, oplog: { acquireCount: { r: 3 } } } 1341ms
[23:42:39] <d-snp> would an oplog query from a secondary block writes until the oplog query result is sent back?
[23:48:33] <d-snp> it could be the oplog queries have nothing to do with our performance problems, they just look a bit suspicious, taking so long
[23:49:06] <d-snp> we collect a bunch of mongodb performance metrics, could they cause oplog queries?
[23:50:19] <d-snp> http://stackoverflow.com/questions/6118411/why-do-my-mongodb-logs-fill-up-with-getmore-local-oplog-rs
[23:50:24] <fxmulder> comparing the count from collection.stats() on the primary and the 'objects cloned so far from collection' count in the new replica's logs, this replica is going to be using a lot more disk space than the other replicas
[23:50:40] <d-snp> apparently its normal for them to take long and they should happen as part of normal things
[23:58:01] <d-snp> if it's really a problem for you, and you can afford some downtime, you can also just restore using rsync
[23:59:09] <d-snp> I'm not 100% certain on what kind of live compaction mongodb does, but it could be that if you run a compact after the restoration is done they'll be closer together