Log file Viewer
#mongodb logs for Thursday the 17th of September, 2015
[06:20:18] <KekSi> does anyone know how to limit the logspam in a cluster? the --quiet option doesn't seem enough
[06:23:11] <KekSi> it looks to me like debug output and not necessarily useful for production (on query routers: balancer round started, ended and did nothing, LockPinger being successful)
[06:23:36] <KekSi> and on config servers a lot of this: 2015-09-17T06:17:26.010+0000 I STORAGE [conn3] CMD fsync: sync:1 lock:0
[08:04:21] <mitereiter> is it just by me or db.runCommand({convertToCapped: "cappedtest", size: 1000, max: 3}) inst working?
[08:05:09] <mitereiter> after the command the collection in collection stats the size and max parameters didnt take up the given values
[09:58:53] <vagelis> If i make query A.B : 123 will it hit the index or i should always use $elemMatch whenever i have dot indexes?
[10:01:49] <mitereiter> is it just by me or db.runCommand({convertToCapped: "cappedtest", size: 1000, max: 3}) inst working?
[10:01:49] <mitereiter> after the command the collection in collection stats the size and max parameters didnt take up the given values
[10:24:43] <joannac> mitereiter: have you read the docs? http://docs.mongodb.org/manual/core/capped-collections/
[10:24:57] <joannac> "If the size field is less than or equal to 4096, then the collection will have a cap of 4096 bytes. Otherwise, MongoDB will raise the provided size to make it an integer multiple of 256."
[10:26:34] <joannac> vagelis: $elemMatch will let you project out a specific array entry. both should use the index though
[10:27:50] <vagelis> joannac well im reading in stackoverflow how u might not hit the index and im trying to filter queries in python that dont have an index and im kinda confused with this elemMatch
[10:44:56] <mitereiter> sorry joannac, I did view the doc page, but i searched for a syntax error not for a limitation
[10:48:48] <mitereiter> is it posible that I cant pass a max parameter by converting an existent collection to capped collection?
[10:57:34] <vagelis> I query an array which has this form: MyArray: [1,2,3..] and i query like that: { MyArray: 2 }...... Shouldnt in explain says that it was multikey? Why it says false?
[11:09:41] <vagelis> MyArray: 2......MyArray: [2].....MyArray: {'$elemMatch': {'$eq': 2} }...........MyArray: {'$elemMatch': {'$eq': [2]} }
[12:00:42] <ronr> hi, I'm trying to setup a mongo shard cluster (mongo 3) and everything went well until I tried starting mongos for the first time. It keeps failing on getting a distributed lock for upgrading the database: 2015-09-17T13:49:03.042+0200 D SHARDING [mongosMain] trying to acquire new distributed lock for configUpgrade on XXXX:27019,XXXX:27019,XXXX:27019 ( lock timeout : 900000, ping interval : 30000, process : XXXX:27017:1442490538:180428
[12:01:03] <ronr> I don't find any errors in the logs of the config servers, any ideas where to look?
[12:03:49] <ronr> but also these (without -v) waited 33s for distributed lock configUpgrade for upgrading config database to new format v6: LockBusy Lock for upgrading config database to new format v6 is taken
[12:11:36] <ronr> I just noticed all config servers claim they're busy upgrading, however they've be doing so for nearly three hours now, on a new empty installation so that can't be right
[12:15:39] <ronr> removing all the data dirs of the config servers and letting them rebuild themselves seemed to do the trick
[12:30:04] <ronr> next issue: if I try to add a replica set as a shard (sh.addShard("rs1/<host>:27017")), it says my host does not belong to replica set rs1, but if I add the host (sh.addShard("<host>:27017")) it says host is part of set rs1, use replica set url format???
[12:36:12] <vagelis> mitereiter: i think i found why its not a multikey. I think that in entire collection has one element in the array. I just inserted one document and put 2 elements in this array made a find and showed finnaly isMultikey:true.
[12:50:12] <vagelis> metereiter actually no, If i search an array that has max 1 element and i use $eq it doesnt work but if i search with $in operator then is a multikey
[12:52:13] <ronr> and just jumping from issue to issue here (is that tutorial really complete?), I managed to get to the next step of my addShard command. But my shard obviously has authentication enabled and therefor mongos can't add it and the addShard function has no means of passing in a username and password? How do I do this?
[13:33:07] <nikitosiusis> I'm trying to import my mongodb bson dump to mongo 3.0.6 via mongos. I set up 256M chunksize, but seems this setting doesn't work for me - chunks don't grow above 64M and new chunks are being created
[13:33:48] <nikitosiusis> my dump starts to import quickly, but after some gigabytes it starts to provide big io and almost hangs
[14:18:44] <deathanchor> find returns a cursor, aggregate returns a single document containing an array with results.
[14:19:27] <bdbaddog> I'm a bit of a n00b. can I store the results from the aggregate in a javascript variable in the shell and use that ina find?
[14:21:08] <pokEarl> If I want to see what requests etc is made in real time I need to use mongosniff, is that right? D:
[14:30:17] <deathanchor> bdbaddog: yes you can do var result = db.mine.aggregate(..).result to store the result array
[14:30:28] <bdbaddog> So I create a set of all referenced _id's between one collection and another, and then want to delete all documents in a second collection which aren't referred to.
[14:32:48] <bdbaddog> > var xyz=db.albums.aggregate({$unwind:"$images"},{$group:{_id:null, photos:{$addToSet:"$images"}}}).result
[14:35:57] <bdbaddog> ugh.. so dumb. I exited the shell and forgot to set the db.. it's working without the .results.
[15:24:59] <kaseano> Hi, I need to use the id of a doc as a FK. I know every doc has _id, using that is a little inconvenient sometimes though, ex I have to convert the string to an "ObjectID" from node, plus passing a small int to the app would have a smaller overhead, vs the long _id string.
[15:25:40] <kaseano> but if I created another ID on top of it, suddenly the BSON would be bigger, plus I'd have to index that new property too
[15:27:56] <Derick> the long _id string isn't actually that long. It's 12 bytes instead of 8 bytes for an int though :P
[16:03:38] <harttho> Anyone know if the mongodb module has known issues with Node4? I get the following weird error when trying to use the module:
[19:07:35] <stickperson> how can i do a query to see if something exists and is not an empty string?
[19:28:25] <BadApe> i dunno if this is true, but i was told if your db is better than 32gb mongo smashes your data?
[19:30:59] <BadApe> i have started playing with the vert.x project, they seem to support mongodb and redis
[19:55:10] <StephenLynx> you sure can follow a strict model for collections on your application, but mongo will NEVER enforce said schema.
[20:05:24] <deathanchor> so... it's only enforced on writes. I hope there will be a way to find all docs which are "not validated"
[20:09:10] <StephenLynx> like, you have a production server and you don't want to take chances with some data being wrecked by a botched patch
[20:10:12] <StephenLynx> or even on a development environment to catch these errors without adding the overhead to the production environment
[20:11:23] <BadApe> orientdb i liked because it had options to have strict schema, partial and schemaless
[20:13:28] <StephenLynx> I dunno, schemas never did anything I couldn't do myself already. I don't hold any love for them. its just one more thing to adjust when I change the model.
[20:14:08] <BadApe> ok teach me something i just did if (document.containsKey("name") && document.containsKey("MaxCharacters")) {
[20:16:09] <moqca> So I'm just discovering MongoDB and I'm trying to port a scheduling sql db I have, that has an employee Table (employee info), dates table (dates prepopulated from now to 10 years, identifying weekdays and weeknums), and scheduleDate table, which takes from employees and dates when an employee is scheduled. Trying to wrap my head around mongodb, how can I do this here?
[20:19:58] <moqca> I don't know, i just find it easier to say, find all the times someone has worked weekends for the past 3 months.
[20:23:13] <moqca> What I'm trying to understand is, how would I do the design in mongo, would the schedule someone is working for this week be an embeddedDocument?
[20:24:17] <moqca> If I wanted to take all leads and workers that belong to X supervisor, how would that be done in a document?
[20:33:50] <cheeser> depends on the query needs, really. it'd be simple enough to query the "worked" collection with the user id.
[21:48:05] <BadApe> i have a number of document that have a configuration type, i would like to reference another document
[21:48:40] <BadApe> yes i know it is a relationship, but i think i need a little bit of structure between documents
[22:07:27] <Fetch> I have a...mongodb dba...who designed an architecture for an app using 3 config instances cohosted with 3 mongos instances (3 datanode mongod replicaset is on a separate set of instances)
[22:07:47] <Fetch> Is that pretty typical with Ops Manager? It doesn't seem to do that configuration for a cluster by default
[22:10:06] <Fetch> joannac: you would not believe the lengths I've gone through to automate installing this beast :\
[22:12:35] <joannac> Fetch: if you open a ticket with your question I'm more than happy to answer on the ticket
[22:14:41] <joannac> and yes, the question should be "Can I get this configuration from Ops Manager" and not "is my dba an idiot" ;)
[22:16:49] <Fetch> (I doubt the dba, who's used ops manager a bit, would give me an architecture that was unimplementable)
[22:22:59] <joannac> Fetch: anyways, thanks for opening a ticket, we'll respond there and continue working with you on the ticket
[22:24:14] <Fetch> I know it's your support interface, I really was looking for a quick discussion on IRC ;)