Log file Viewer
#mongodb logs for Thursday the 23rd of May, 2013
[00:14:37] <kfox1111> Is there a way to get an insert to a collection in mongodb to never create the collection if it is dropped?
[00:18:47] <kfox1111> I have another one that I am having drop the collection. When dropped, I want the first process to fail insert, notice something happened, and choose the new collection.
[00:19:34] <kfox1111> Long story short, I'm using capped collections, and need an easy mechanism to allow me to drop and recreate them with different sizes/etc and have the watchers pick up on it.
[00:23:23] <kfox1111> hmm.. so what I am thinking currently is, I have an "init" command that creates my capped collections. I have a process that listens on http and inserts into that collection, and I have a set of watchers that tail the collection.
[00:23:41] <kfox1111> When re'inited, it bumps a version, creates a new collection, and drops the old one.
[00:24:44] <kfox1111> the thing doing the inserting should notice the collection is gone and start writing to the new one. but it ends up recreating the old collection, and not even creating it as a capped collection causing things to misbehave.
[00:25:17] <kfox1111> a readWrite role wouldn't help in that case either, since it is the one doing the writing.
[00:28:01] <kfox1111> There is some unique id's involved that need to be reset when rolling to the next collection.
[00:29:04] <kfox1111> The problem I'm trying to deal with is if I don't drop the collection, then the tailers will tail forever on a never going to change collection.
[00:29:49] <kfox1111> Once dropped, then when they try to recover, they notice the new db version and start tailing the correct collection.
[00:31:04] <kfox1111> The good news is, if I ensure the thing doing the insert is not running while I run the init program, it does work as expected, so I can work around it for now.
[00:31:49] <kfox1111> There still is a lacking feature there though where an insert/update needs to support a mode where collection creation is nonautomatic.
[03:53:55] <MasterGberry> How does the data storage work exactly? I am interested in knowhing why mongodb db's take 192mb when first created?
[03:59:04] <Derick> which reduces the file sizes for initial init, and noprealloc prevents preallocation
[06:46:01] <seban> Hi, what does it mean that I have document with "_id" : "configUpgrade" in "state": 1 in my config.lock?
[07:45:18] <seban> Hi, what does it mean that I have document with "_id" : "configUpgrade" in "state": 1 in my config.lock?
[07:52:50] <crudson> tagrudev: http://docs.mongodb.org/manual/reference/operator/rename/#op._S_rename
[07:54:53] <crudson> introduced in 1.7.2 according to docs, but I see in 2.2 some behavior was changed, so I'd review that page for expected results.
[07:56:13] <crudson> tagrudev: that is implied by my previous comment, yes. Give it a whirl and see.
[08:27:36] <igungor> joannac: i've prepared a test case, its results with mongod logs and all, but the problem was that i should be using elemMatch
[08:43:41] <someprimetime> should I create an array of children comments inside one parent comment in my comments collection?
[08:44:20] <someprimetime> or create a new entry for each record and just mark them as having `parent_id` !== null
[08:44:38] <someprimetime> if not null then it'd be referencing another record which would be its parent
[08:58:47] <crudson> someprimetime: if comments can be indefinitely nested then keep them as top-level entities and link to parent. querying will be a headache otherwise.
[09:24:25] <crudson> uh huh. I just moved to the US pacific north west and the wildlife here is a bit weird. The noises these things make...
[09:26:14] <crudson> but nothing like hearing vixens shriek at night when I lived in the English countryside...chilling
[09:42:10] <anton123> I have a collection users with telephone key, in this there are numeric values also string values. When I try to find any number which is numeric in the collection it does not show but the string values are coming in the search.
[09:43:25] <Zelest> any telnet that doesn't swallow my ^C should die in a fire.. no matter how cool they are :(
[09:51:02] <anton123> I am using mongodb with php and this is the array I am passing: $column='telephone';$collection->find(array($column => '9899980668'));
[09:58:43] <anton123> values are stored both in string and numeric format in the key telephone while i did mongoimport from mysql. those which had - or + were converted into string and others as numbers
[10:02:10] <anton123> actually there is an input box where I will pass something like 9898 and it should show both the numeric and string entries. I used regular expression /searchterm$/
[10:07:21] <anton123> in telephone field if i search only 9 it shows me all the entries which are in string value having 9 in it. where as the phone numbers which are numeric and contain 9 doesnt show up
[10:33:27] <Hounddog> hi there, what would be the best way to store json to a specified column? array or convert the json t string?
[10:35:05] <Hounddog> well i am recieving tweets and items from instagram so i have 3 fields, type, item_id and description in the description i just want to save to complete json recieved from twitter or instagram
[10:37:46] <Hounddog> well not really, just have to send the json later back to the client but that can be a conversion also
[11:18:09] <remonvv> Does anyone know how mongos manages its connections? Meaning, if I make X connections to mongos will it make X connections to each shard resulting in SHARD_COUNT * X total connections on the mongos instance?
[11:18:35] <remonvv> Or will it start reusing connections at some point. I'm having trouble finding info on this.
[11:19:10] <spolu> hi, when running a find with readPreference: secondary, I see duplicate records coming from chunk that were recently migrated. When running the find on the primaries (no readPreference) the document appears only once. Anybody has a clue as to why?
[11:22:09] <remonvv> spolu, perhaps the scenario where the secondary of the old shard still has a copy because it hasn't reached the migrate in the oplog yet and the secondary of the new shard already processed the migration.
[11:23:48] <remonvv> No idea, that's speculation at best. I'd normally assume this wouldn't be possible (since it allows to return 2 documents with the same _id for example, which is the case here I assume).
[11:25:15] <spolu> remonvv: no clue anything between a few hours to 2days (added a shard to the cluster)
[11:25:53] <spolu> remonvv: I know the value of the shard key of at least one of the culprit document
[11:41:30] <quattro> i need more read capacity on one of the replica sets since there are some other collections on it that need more reads, can i just make it 5 replica's on one and 3 on the other?
[12:09:19] <Snebjorn> will $match take advantage of indexes after an $unwind? For example db.text.aggregate({$unwind: ...}, {$match: ...})
[12:24:47] <binnyg> I am unable to do a query with java driver. fineOne() works fine but find with a query object is throwing an error(java.lang.IllegalArgumentException: can't serialize class com.google.common.base.Present).
[12:31:19] <binnyg> may be I wasn't clear. I have a collection and I am trying to search the collection. I started with findOne() and can see the result. When I try to use find on collection with a query I get the above error.
[12:33:10] <remonvv> There's no way a findOne() wont throw an error where find() will with the exact same query.
[12:44:25] <binnyg> name is a google Optional class which I totally missed. It works. Thanks remonvv
[13:09:37] <fatih> I have field named "foo" with lots of subfields. I can test a subfield via {"foo.subfield" : {"$exists" : true}}, however {"foo.sub.field" : {"$exists" : true}} doesn't work
[13:44:27] <fatih> I've tried anything Nodex, for queryin I'm using monoghub (a gui client on mac osx 10.8)
[13:44:51] <fatih> now I'm changing my codebase (data structure) that the subfields will be not keys anymore
[13:50:50] <fatih> ok when I have an array with items like {"name":"foo", "date":"123"}, {"name":"bar", "data":"124"}
[15:20:09] <remonvv> How is it that wherever I am in the world I get periodically disconnected for no apparent reason.
[15:22:08] <remonvv> You are one less now, possibly. Unless you were accessed from a different thread.
[16:16:40] <LzrdKing> is there a way to retreive detailed statistics about a database from the mongod http interface? like what http://127.0.0.1:28019/listDatabases?text=1 provides, only more information than just sizeOnDisk, like number of collections, indexes, objects, etc
[17:07:48] <yann_ck> Hello Guys, Is it possible to aggregate items, group them to get a min and output the corresponding document id?
[17:11:37] <LzrdKing> is there a way to retreive detailed statistics about a database from the mongod http interface? like what http://127.0.0.1:28019/listDatabases?text=1 provides, only more information than just sizeOnDisk, like number of collections, indexes, objects, etc
[17:20:14] <yann_ck> I have something like Item.collection.aggregate({"$project" => { _id: 1, user_id: 1, created_at: 1}}, {"$group" => {:_id => "$user_id", "min" => {"$min" => "$created_at"}}})
[17:22:07] <yann_ck> and that output exception: the group aggregate field 'item_id' must be defined as an expression inside an object
[18:08:00] <Gargoyle> LzrdKing: About as much as you know about google it seems! http://docs.mongodb.org/ecosystem/tools/http-interfaces/
[18:15:24] <jaCen915> Hi, does anyone know how to query for documents using $where and $in AND sorting those results?
[18:21:34] <kali> jaCen915: it also used to raise huge concurrency problems (only one js interpreter per process) but that bit is supposed to be solved since 2.4
[18:22:20] <kali> jaCen915: $in will be quite fast, and the index will be able to optimize it and restrict the number of docs to pipe through the $where
[18:23:31] <jaCen915> kali: are you saying where is okay to use if restricting the doc results with $in or that $in by itself is good enough
[18:24:26] <kali> jaCen915: it depends what you do with the $where... $where allow to express virtually any condition whereas the native operators have a finite featureset
[18:26:07] <kali> i guess this is mongoose translated, i would assume it generates currentStatus: { $in : <searchStatue> }
[18:56:37] <orospakr> hey, I'm having a bit of trouble using the fancier query selectors when nesting: https://gist.github.com/orospakr/5638498 (I assume I've probably made a goofy mistake)
[20:04:52] <LzrdKing> Gargoyle: i was thinking more along the lines of http://docs.mongodb.org/manual/reference/command/nav-diagnostic/
[20:05:22] <LzrdKing> most of those commands work over http, but very few give me info about the individual databases
[20:07:02] <Gargoyle> LzrdKing: Well, if the built in interface isn't up to mustard for your needs, that leaves you one solution...
[20:07:38] <LzrdKing> yeah, learn php and hack up this script that gets the data i want , but not in a way i can use
[20:08:55] <LzrdKing> yes, except that if i learn pythion, this php scrip that does 90% of what i need won't help me at all
[20:38:58] <LzrdKing> wtf? somehow i created empty databases called collStats, connPoolStats, and listCommands just by curling some urls
[20:52:40] <ThePrimeMedian> can someone help me with a query? I am using mongoose.js (for node.js) here's my schema: http://pastebin.com/VKn2Q7Z3 but what I need is to find all Quest documents that match challenges.metric='user.sigin' and challenges.required >= 5
[20:52:58] <orospakr> is it possible to use $elemMatch on an array of strings rather than an array of objects? I'd like to use the $ operator and while using the flat { myarray : "value_to_match_in_array" } matches, the positional operator is not aware of it.
[20:53:49] <orospakr> (my goal is to replace a string in an array with a different string in a single update operation, if I can)
[21:05:57] <orospakr> (if given a string to match, $elemMatch complains that the parameter is invalid)
[21:16:48] <ThePrimeMedian> can someone help me with a query? I am using mongoose.js (for node.js) here's my schema: http://pastebin.com/VKn2Q7Z3 but what I need is to find all Quest documents that match challenges.metric='user.sigin' and challenges.required >= 5
[21:25:06] <ThePrimeMedian> jaCen915: I can connect to mongodb; mongoose is just an ODM ontop of Mongoose. So if we can figure out howe to make this query, I can adjust it for Mongoose
[21:26:23] <jaCen915> ThePrimeMedian: cool…I'm using mongoose right now, let me play around real quick, for now look up the mongo "find" and "gte"
[21:27:22] <ThePrimeMedian> jaCen915: cool. I know how to use MongoDB and MongooseJS - but I have never tried to query an object by it's name inside an array inside a document. and get the document..... know what I mean?
[21:27:52] <jaCen915> ThePrimeMedian: exactly :) i haven't either, I want to see if my ideas will work
[23:33:37] <iNick> Hi guys. We have a realization of someone else's stupid idea. We have 3x servers in a replica set and the hosts entered for the replicaset are static private nonrouteable IPs. Where is the replicaset info stored, since we only backup one database daily
[23:46:07] <iNick> redsand: maybe I should rephrase... is replset info in the master db, in the production db we're backing up..?