Log file Viewer
#mongodb logs for Tuesday the 2nd of October, 2012
[00:38:57] <acidjazz> can someone tell me why this small snippet of code fails to produce an image when is_file is false on line 12? http://pastebin.com/WwAQCWBW
[00:59:28] <acidjazz> here it is simplified, http://pastebin.com/KrJvdEhH, like i said when line 12 is false (is_file) and it has to write.. the browser says "invalid image" .. if i refresh.. works fine. 0 php errors/notices/etc 0 apache errors and so on..
[01:29:02] <Skillachie> I was wondering if someone could point out what is missing from my map-reduce function
[02:11:10] <Antiarc> I just upgraded my replica set from 2.0.4 to 2.2.0. 3 nodes, the 2 slaves upgraded just fine. I stepped down the master, verified that one of the slaves had become primary, and shutdown the primary to do the upgrade.
[02:11:58] <Antiarc> However, I re-checked my secondary and rs.status() was reporting two secondaries, one down, and no master.
[02:12:30] <Antiarc> Once my formerly-primary node finished resyncing and came back to available, it resumed as primary and things seem to be fine. There didn't appear to be any interruption of service, but rs.status() reporting no primary makes me mighty nervous.
[02:14:09] <Antiarc> I guessed tied election, but A) there was already a primary doing fine, and B) I'm not sure what to do about that; the doc advises against having an arbiter in replica sets with an odd number of members, which mine does. Should I bring an arbiter up next time I'm doing upgrades, so there are 3 nodes up and voting at all times?
[02:14:29] <Antiarc> (I was unable to add an arbiter to the set to cast a tiebreaking vote, as there was no primary to update the rs config on!)
[02:17:51] <Antiarc> Nevermind, reporting tools show an interruption of service. So yeah, that's bad.
[02:22:46] <Antiarc> Then, it shut down, leaving the RS without a master, and 2 nodes up that deadlocked on election.
[02:23:28] <Antiarc> So my question is, why do the docs say to NOT run an arbiter in a cluster with an odd number of nodes? Arbiters seem like they should only be used for tiebreaking, and when I lose a node in a 3-node cluster, I am down to 2 nodes and need an arbiter, but I don't want one for normal operations since that'd be four votes in the pool.
[03:20:36] <owen1> I use the Mongo DB Native NodeJS Driver - chains.update({x: data.id}, data, {safe: true, upsert: true}, function(err, chain) {}); but it's not upsertng and i have collections with the same x. when i try this from the console it's working as expected - db.chains.update({'x': 2},{'x': 2, 'y': 5}, true) any idea why?
[08:54:11] <gyre007> can I delete local/journal files from mongo live while mongo is running ? Should I not delete them at all ?
[09:11:34] <nickbr> Hi, I'm having problems with the Mongo PHP driver, connecting to a replica set. I'm intermittently experiencing connection exceptions: couldn't determine master. I've seen posts about, like this one: https://groups.google.com/forum/?fromgroups=#!topic/mongodb-user/CPhQmpTW4GA
[09:11:46] <nickbr> However, I have tried closing the connection after performing a query, then performing the query again, and it still fails
[09:12:36] <nickbr> Most of the time it works well, but every so often it starts throwing exceptions and only works about 1/20 attempts
[09:17:08] <gyre007> its just that we are running out of disk space at horrible misallocated VM ....and the most of it is occupied by journal and local mongo files
[09:17:10] <wereHamster> gyre007: better not delete any files. Why are you intentionally trying to fsck up mongo?
[09:18:21] <Derick> nickbr: any change you can try the latest code of github for the driver? we've rewritten loads of it
[09:21:36] <nickbr> I tried 1.3.0beta 2 last night, does that count? instead I just kept getting the error 'No candidate servers found'
[09:24:18] <LoonaTick> If I execute "db.things.remove({});", will mongodb clear up disk space as well?
[09:34:12] <ppetermann> gyre007: http://stackoverflow.com/questions/4555938/auto-compact-the-deleted-space-in-mongodb/4560096#4560096
[09:38:01] <wereHamster> gyre007: files are where mongodb stores its data. If you delete the files, you are going to have a really bad time.
[09:39:29] <nickbr> Derick: I tried 1.3.0beta 2 last night, does that count? instead I just kept getting the error 'No candidate servers found'
[09:42:37] <nickbr> also tried including the other servers: mongodb://phpuser:[pass]@server1,server2,server3
[09:52:11] <Derick> nickbr: turn on MongoLog and it will tell you exactly what happens (with: MongoLog::setLevel( MongoLog::ALL ); MongoLog::setModule( MongoLog::ALL );
[09:53:08] <Derick> but do try latest github, as we've fixed things since 1.3.0beta2 regarding auth and replicasets
[09:53:29] <nickbr> Derick: Will do. I've got the driver installed via pecl, how should I go about installing the git version?
[09:54:24] <Derick> git clone https://github.com/mongodb/mongo-php-driver && cd mongo-php-driver && phpize && ./configure && make && make install
[09:55:35] <nickbr> Derick: great, thanks. If I want to revert to 1.2.12, can I just do pecl install -f mongo-1.2.12
[11:00:56] <LoonaTick> Hi everybody. I've ran 'db.dropDatabase();' on all databases I have, but disk space is only freed half. Anybody know what to do?
[11:30:05] <dforce> I did a find() and a for loop to show some elements on a webpage (with this no problem)
[11:40:03] <Derick> MongoDBIdiot: also, IRC etiquette dictates that you don't argue with ops. So drop it.
[11:41:46] <Derick> dforce: if you have code, please put it in a pastebin (pastebin.com or something similar)- don't paste it into the channel please
[11:49:31] <dforce> in django I receive error message 'Variables and attributes may not begin with underscores: 'junk._id' '
[11:50:34] <Derick> dforce: junk['_id'] perhaps? I am sure there is an example on the internet somewhere
[11:52:35] <Derick> dforce: http://villaroad.com/2010/02/return-mongodb-ids-inside-django/ seems what you want to read
[12:05:01] <iksik> I'm trying to understand how replica sets works... and for now, i have some problems wih understanding figure number 4: http://goo.gl/KTHMJ ... how (for example) python client will know, to which hostname he can connect after first master failure?
[12:07:50] <kali> iksik: the driver grabs the replica layout when it opens the connection, but you can also start it with the names of the three servers
[12:15:20] <Skillachie> Hi I am trying to complete a map reduce job in Mongo. I would like to aggregate based on if a path is present example
[12:15:39] <Skillachie> db.file_vault.group( { key: {"path": true}, initial: {sum: 0}, condition: {"path":/\/Users\/mongouser\/Downloads/} , reduce: function(doc,prev) {prev.sum += doc.size} })
[12:16:08] <Skillachie> How do I implement this in map- reduce ? Complete description of problem here http://pastie.org/4894244
[12:17:11] <kali> Skillachie: first of all, the value emit()ed should be of the same form than the return value of the reducer
[12:19:01] <Skillachie> ok. So how do i specify that the path in the emit should be aggregated if the path is within the path ? (hope this doe not sound confusing)
[12:20:33] <kali> Skillachie: you need to call emit for each bit actualy: for "a/b/c" you'll call emit("a", { size: sizeof_abc}), emit("a/b", <same value>), emit("a/b/c", <same value again>)
[12:23:30] <Touillettes> Do you know if it's possible to use the continueOnError flag for a batch insert with perl ?
[12:26:48] <Touillettes> Do you know if it's possible to use the continueOnError flag for a batch insert with perl ?
[12:30:19] <Touillettes> Do you know if it's possible to use the continueOnError flag for a batch insert with perl ?
[12:50:10] <Derick> http://mirror.umd.edu/roswiki/doc/electric/api/mongodb/html/rs__member_8h_source.html has all the states for future reference :-)
[12:57:01] <Touillettes> Do you know if it's possible to use the continueOnError flag for a batch insert with perl ?
[13:04:48] <schlitzer|freihe> hey all, can someone tell some details why it is needed to shutdown a replicaset while updating from 2.0 -> 2.2 when authentication is enabled?
[13:06:52] <schlitzer|freihe> in a sharded setup, can i still switch config & mongos to 2.2 & have 2.0 shards with autentication enabled in the whole cluster?
[13:44:38] <oskie> anyone seen this: [rsSync] replSet syncTail: 13297 db already exists with different case other: [test] me [Test], syncing: { ts: Timestamp 1348427813000|1, h: -3983475495887886658, op: "c", ns: "Test.$cmd", o: { dropDatabase: 1 } }
[14:52:05] <schlitzer|freihe> didnt notice that when i read it changelog the first time, seems i will simply wait for 2.2.1 to come up
[15:19:56] <LoonaTick> To clear up disk usage I was suggested here to clear up the data directories. Afterwards, I accidently created shards instead of sharded replicate sets. I now want to remove these shards again, but it complaints the slave is not ok and therefor some databases cannot be removed, and therefor not all shards can be removed. What should I do?
[15:20:55] <LoonaTick> Sorry for the most likely stupid question, I took over mongodb administration from a colleague, I have no experience with it but found out the data in there isn't correct. I decided to completely rebuild the data, but after dropping all databases the disk space wasn't cleared.
[15:26:28] <_m> Let me know if that works. I've never experienced said behaviour *and* dropped the collection.
[15:27:07] <LoonaTick> _m: I already completely dropped the data files, setting it up from scratch (also made this decision to understand the inner workings better, by the way)
[15:29:56] <LoonaTick> _m: Thanks, but I don't think that applies to my previous situation, as I did both drop and repair. Perhaps there were some other databases though...
[15:34:33] <_m> Without more information, I would assume you had memory free to mongo but not the filesystem
[16:13:38] <LoonaTick> db.runCommand( { addshard: "rs_b/nosql1.backend.tc3:27022,nosql2.backend.tc3:27022,nosql3.backend.tc3:27022" } ); --> "can't add shard rs_b/nosql1.backend.tc3:27022,nosql2.backend.tc3:27022,nosql3.backend.tc3:27022 because a local database 'gpinternationalAvatars' exists in another rs_a:rs_a/nosql1.backend.tc3:27021,nosql2.backend.tc3:27021,nosql3.backend.tc3:27021"
[16:14:02] <LoonaTick> I have just enabled gpinternationalAvatars to be a sharded database (I have 1 shard) - should I wait for something to be initialized?
[16:22:21] <LoonaTick> Ah never mind, I just drop the databases and they can be added again (some script adds these databases)
[17:37:41] <dforce> I am using pymongo and django and want to see the ObjectId. Derick pointed me to http://villaroad.com/2010/02/return-mongodb-ids-inside-django/ but I get an error within the python script. Can someone help me solve this problem ?
[18:35:13] <jmpf> sometimes mongodump will fail http://pastie.org/private/l9tlr2noe1mrv9vled2gha <--- but `df -h` shows plenty of space on both the server and this remote host that we are dumping too; I do note that virt mem is @ 64.5G && total mem is 66G
[18:50:19] <JmZ> installed via the gen10 repo on centos5. why does it feel the need to allocate 3GB+ of space for 'journal' files?
[18:51:02] <JmZ> i can't start it because there isn't 3GB of free space, why would it need that much? It is a clean install, there's no need for using such an amount of storage. am i missing something? maybe a setting?
[18:53:46] <_m> JmZ: You can set the jounral size to be smaller by using the command line "--smallfiles"
[18:56:45] <JmZ> _m: the error its self says to use --smallfiles, but not *why*. so less 'look harder next time' implications please. came here for clarification
[19:02:18] <_m> As that's a 44-minute presentation, I haven't filtered it for your answer. However, " we'll give an overview of durability with MongoDB, demo journaling, and discuss journaling internals."
[19:15:37] <leehambley> is there such a thing as a blocking findAndModify for MongoDB? I'm trying to re-write this queue implementation backed by MongoDB https://gist.github.com/25258909667d3b5ff75a
[19:38:47] <alaska> If I have a collection indexed by a unique string say "name", and I want to do concurrency checks on a revision key "_rev" does it help me to index _rev even though all updates will refer to both the primary key and the _rev?
[19:40:47] <alaska> Basically, I'm ensuring that I know if someone has touched an entry since I pulled it from the server. When I go to update, I update based off of both {name: 'name I pulled', _rev: 'rev I pulled'} and I $inc _rev. That way, if the rev has changed in the meantime, it won't update and I know it's stale.
[19:58:19] <KyleG> So I've searched and searched and it seems like it may be a new issue limited to 2.2….I'm seeing the following in all my secondary nodes mongo logs: Tue Oct 2 12:49:03 [rsSync] info DFM::findAll(): extent e:3f51ee00 was empty, skipping ahead. ns:local.replset.minvalid
[19:58:39] <KyleG> and then the following in my cfg server logs: Tue Oct 2 12:56:28 [conn11002813] problem detected during query over config.locks : { $err: "Invalid BSONObj size: 0 (0x00000000) first element: EOO", code: 10334 } which follows a generated stack trace
[21:55:43] <systemizer> anyone have an issue when a replica gets into a rollback state, but it doesn't create a rollback directory in the dbpath?
[22:22:40] <timeturner> at first I wasn't embedding the comments at all and I was simply referencing the post's ID in each comment. comments and posts were stored in separate collections
[22:23:26] <timeturner> and finally I thought about embedding the whole comment documents into an array in the post
[22:39:04] <Skillachie> kali: got it solved thanks for pointing me in the right direction this morning
[23:33:27] <Killswitch> Hey guys, I did mongodump, and I was wondering is it possible to just import a single bson file into an existing collection using mongorestore?
[23:59:18] <scoates> more findandmodify problems. )-: Tue Oct 2 23:57:07 [conn9412980] warning: ClientCursor::yield can't unlock b/c of recursive lock ns: gimmebar.jobqueue