Log file Viewer
#mongodb logs for Tuesday the 3rd of November, 2015
[00:10:45] <daidoji1> hello, in pymongo is it possible to have multiple '$addToSet' operations in an update?
[00:11:06] <daidoji1> currently its messing up because those 'addToSet' strings are treated as keys in a python dict
[01:42:51] <sabrehagen> i'd liek to run db.currentOp() to find long running tasks. this blog post suggests i need to be an admin user: http://blog.mongolab.com/2014/02/mongodb-currentop-killop/
[01:43:38] <sabrehagen> i've authed against my admin daatabase and my role is 'userAdminAnyDatabase' but runnign db.currentOp() returns { err : 'unauthorized' }
[02:53:28] <Boomtime> in short, userAdminAnyDatabase is a role for administering users, it can't do much in the way of data or realtime control type access
[02:53:58] <Boomtime> however, that role does permit you to either grant yourself the extra needed privileges, or to grant somebody else such privileges
[02:55:22] <Boomtime> the most restrictive built-in role that includes the ability to currentOp is clusterMonitor
[02:56:05] <Boomtime> if you don't want all the extra stuff that comes with, you can just grant yourself the specific privilege to issue the 'inprog' action (which is the real command behind currentOp)
[03:23:03] <sabrehagen> also, in 2.6 .explain() returned a very different result to explain() in 3.0. i've read about the changes on mongodb.org, but can't find out how to run the same style of explain() as 2.6. i'd like to know the cursor type and nscanned* stats in 3.0
[04:23:50] <cppking> hello guys , in a 3 nodes repl-set , 1 master and 3 secondary, if one secondary is down , at the meantime, master failed, will this repl set still work normally?
[04:25:56] <Boomtime> "in a 3 nodes repl-set , 1 master and 3 secondary" <- are there 3 nodes, or 4?
[04:26:54] <Boomtime> in either case, if 2 members fail, then you don't have a majority - also, avoid the term 'master' as that was used in the old system, the correct term is primary
[04:31:58] <cppking> one secondary is down , then primary failed , will this replset still work normally
[06:03:36] <cppking> joannac: what if 1 primary and 3 secondary , primary failed when one secondary is down , after election , while this repl still work normally?
[06:04:26] <sabrehagen> in 2.6 .explain() returned a very different result to explain() in 3.0. i've read about the changes on mongodb.org, but can't find out how to run the same style of explain() as 2.6. i'd like to know the cursor type and nscanned* stats in 3.0
[06:11:15] <btcbuy314> I'm making a music playlist website and I have two tables, one which stores songs, and the other, playlists, which stores title of playlist and an array of _id of different songs
[06:12:04] <btcbuy314> how would I do this? Everything seems to have no errors, but then the database table is unchanged
[06:27:50] <phreelynx> @btcbuy314: please elaborate "I'm having a lot of trouble updating the order of the array"
[06:41:05] <btcbuy314> so basically, i have an array of ids saved in the playlist table, I iterated through the array, changed each one to the next one and the last to one before
[06:44:39] <phreelynx> how about having a separate key, in the playlist documento, for the current_song which contains the index of the song in the playlist array. Once completed you can just change current_song into index+1
[06:46:27] <btcbuy314> im not actually changing the order to +1 , i just said that to make it simpler
[06:50:56] <btcbuy314> heres a pastebin, the msg.indexList is basically a list the same length as doc.videos but whats in it would be something like [0,2,1,4]
[06:53:39] <btcbuy314> something that I think will help solve this is doc.videos.pop() sucessfully deletes the last video, but if I try to duplicate the first video with doc.videos.push(doc.videos[0])
[06:54:05] <Silverfall> if I have 100000 documents and I want to check the one with the highest "{ size: ...}" value, does limiting search results increase my performance or it's useless since mongo has to iterate over the whole collection anyway
[06:55:01] <preaction> Silverfall: and if you find yourself making that query often, you might want an index
[07:02:49] <Silverfall> preaction so even if I already have a running database full of "size" values, entering this single command in the shell "db.mycollection.createIndex( { size: 1 } )" would allow me to permantently keep them ordered by size without making mongo work too much?
[07:03:35] <preaction> it creates an index that's ordered by size. as for how much work happens, well, that's the trade-off. the index must be updated when a write happens
[07:04:40] <Silverfall> so unless I have an amount of X documents it isn't worth to make the index to improve overall performance.. could you define the value of X in this case?
[07:05:44] <preaction> there are too many variables for me to even fathom a guess at what one of them, X, could be
[07:06:17] <Silverfall> well, I see two variables involved: the frequency I write new documents and the frequency I check the whole collection ordered by size
[07:07:02] <preaction> you're missing: the amount of memory used to hold indexes. the size of the documents that must be scanned. the speed of the disk used for storage.
[07:07:32] <Silverfall> ahahah, I think I'm just gonna leave it like it is now.. if it starts to slow down I'll apply an index XD
[07:07:50] <preaction> or just whip up a dev environment, which you should already have, and run some benchmarks
[07:09:42] <cppking> In mongo 2.x , Does every insert or update action will raise an implicit lock which is database level??
[07:09:44] <Silverfall> there are too many variables, like user behaviour, that I cannot predict programatically
[07:28:13] <cppking> In mongo 2.x , Does every insert or update action will get an implicit lock which is database level?
[07:29:08] <nishu-tryinghard> guys i made dump of a remote db and i want to restore it locally and use it.
[07:31:55] <nishu-tryinghard> cppking, nvm i was just new so i thought dump will directly make a db locally. I came to know that i need to use restore as well to make a copy from the dump.
[07:33:06] <nishu-tryinghard> cppking, can i use mongorestore without --dbpath param so that it can use the default one?
[07:55:21] <Iskandar> how do i specify witin the mongo.conf that it should be run a s a configserver?
[07:56:55] <Iskandar> manual says that i should run the mongod from the command line, but i want to use the mongod as a service using the conf file
[08:39:52] <Iskandar> hmmm, the yaml conf fails with redhat service mongo init script: dirname: missing operand
[08:57:10] <sabrehagen> how do i tell what type of cursor a query is using with mongodb 3.0? it's not shown when using .explain() for me
[09:16:16] <Iskandar> how do i check if my conf is valid ? i get a failed but cant find where the conf error is
[09:17:46] <csst0111> Somehow i messed around with my db and trying to re-import the json files gives me : "error inserting documents: E11000 duplicate key error index: ..."
[09:20:52] <Mindiell> hi there, is the difference between distrib packages and .deb packages is really a problem ?
[09:25:49] <Iskandar> btw, on redhat im missing the mongos service in /etc/rc.d/init.d since i want ot run a mongos and config server on one machine for testing purposes
[09:30:55] <cppking> Iskandar: just copy /etc/rc.d/init.d/mongod to /etc/rc.d/init.d/mongos then do some change on it
[09:39:58] <Iskandar> sort of, i have 3 machines, i want to use 1 machine for the router and configdb and the other 2 as shards
[09:42:11] <cppking> you need to install another mongodb server in seperate dir like /opt/ , then boot it up
[09:43:34] <cppking> so you need to install another mongodb-server from tarbal, then configurate is as a router server
[09:49:02] <cppking> maybe you don't need to install another mongo, just start mongs with correct configured options like --dbpath and --shardingRole
[10:20:40] <Mindiell> is it the same thing to have two collections in one database and two databases with one collection in each ?
[10:33:38] <Iskandar> i have made this script: http://pastebin.com/THmUJj1E when i run mongos --config /etc/mongos.conf the mongos runs fine and dandy, but when running service mongs start, i get a fail. The mongos.log leaves no trace for me
[11:14:34] <Iskandar> http://pastebin.com/cnJkA8KA here is my output using the same /etc/mongos.conf
[13:59:26] <deathanchor> CustosLimen: depends how big the data is on your configsvr, but usually a minute
[13:59:41] <deathanchor> doesn't affect anything while it goes on, cause you should have your balancer off
[14:04:10] <CustosLimen> deathanchor, I was actually not clear on what exactly was effected - when I realized its only config server data I stopped worrying
[14:11:10] <ramnes> does anyone know if there is a way to tell PyMongo to give the hand back directly when doing a collection.create_index()?
[14:26:10] <Iskandar> if i use rs.addshard(), will this be persisted so when i reboot the server, this mongo instance will be part of the shard again?
[14:31:58] <yopp> will aggregation framework block something, when result is written to the output collection?
[14:35:26] <pamp> its a standalone server.. I will have problems if I stop the process before completion?
[14:36:11] <cheeser> yopp: yes. and there will be read locks, such as they are, on the input collection
[14:39:05] <yopp> cheeser, so pretty much the same as with map/reduce with output to the collection?
[14:45:39] <yopp> cheeser, to simplify. I've got few sharded collections for pre-aggregated data (raw data, 1h, 1d, 1y & etc). Aggregation is made step by step, from raw to 1y collection. Is there any benefits in using aggregation framework instead of map/reduce in this case?
[14:48:10] <cheeser> aggregation will be more efficient OOB since it can work with the bson directly.
[14:49:43] <yopp> So, if right now with map/reduce we're in fact running only one map/reduce at the time, with aggregation they can run simultaneously? With document-level locking, only affected documents will be locked, right?
[14:52:03] <yopp> Apart of having 100mb limit, is there any other "catches" with aggregation framework?
[15:07:32] <repxxl> hello, i have 2 fields like "name" : "This is a headline", "tags": ["School", "5B", "2015"] i wanna search by name and the same time also search array values and return the matching with a priority to show tag founds first then name
[15:08:15] <repxxl> what do i need exactly ? i was thinking something like regex "name" $or and $in together
[15:27:14] <repxxl> cheeser i wanna search for single field or array values giving a priority to return first array values in mongo cursor
[15:36:27] <ruphin> Hi, I have some questions about upgrading my config servers to 3.2. I'd like to use the new config server setup with replica sets, since I am having issues with config servers running out of sync. I haven't found a manual anywhere.
[15:38:59] <ruphin> Can anyone here give me a few quick pointers on how to migrate? I have some experience working with mongo, just need to know what the basic strategy would be
[15:39:55] <ruphin> I'm thinking it would probably be easiest to isolate a single config server in my current setup, restart it with 3.2 and configure it as a replica set member, then add new empty servers to that replica set
[15:40:51] <ruphin> If someone can confirm that this is safe, I can go ahead with the above procedure
[15:43:18] <repxxl> cheeser but its not working for me can you maybe tell me what is my problem ? or how to do it right ty
[15:53:45] <repxxl> cheeser ok however it works how i can sort now the find query to get the matched "tags" first instead of "name" matches
[15:55:15] <repxxl> cheeser the appropirate documents i mean so i get a list of all found "name" and "tags" but get the tags found documents first in the cursor
[16:06:32] <x1nux> Hello ! I have a mongo DB 2.6.7, and my hard drive is full, I deleted some records, but the disk size didn't go down, is there a way to delete those records to reduce disk size ?
[16:08:28] <xhip> quick question: since my apt-get dont want me to update my mongo.. how can I do it manual?
[16:13:32] <x1nux> cheeser, but this command "db.repairDatabase()" needs free space, in harddisk or not ?
[16:16:39] <cheeser> bring up a new replset member with more space. the intial sync will end up using less space anyway.
[17:27:27] <Jacoby6000> If i have a collection with a.b.c.foo and d.e.f.foo, how can I set all keys "foo" to a value?
[17:27:47] <Jacoby6000> basically I want to set a key to a certain value, regardless of where it is in the tree
[17:35:10] <symbol> Jacoby6000: As far as I know, that isn't possible within Mongo. You'd need to handle that server side or structure your data in a different manner.
[18:21:04] <symbol> Not sure what you mean by rolling upgrades. I'm not currently using wiredtiger. Probably should be though.