Log file Viewer
#mongodb logs for Friday the 6th of December, 2013
[01:54:36] <iwantoski> why can't i do db.collection.findOne({_id: "52a108be95687e0e6184e07f"}, function(err, item) { .. }) ?
[04:27:02] <a|3x> what could be wrong with my text search attempts? http://pastebin.com/GHkDbfP9 why is it not finding my document in both cases?
[07:59:55] <spicewiesel> please, what's the best way to copy the data from on sharded setup (mongos+cfgsrv+replicaset) to a similar test environment? (same setup)
[08:45:32] <Seraf> hello everyone. I have a replicaset. All nodes are stuck in STARTUP2 mode ... I know which one should be the primary, how can I force it please ?
[09:10:35] <liviud> I have several nodes with 2.2.0 mongodb. If I upgrade some to 2.4 and use the 2.4 drivers in a replicaset configuration will the 2.4 drivers be able to talk to the 2.2.0 mongodbs ?
[09:16:09] <kali> liviud: check out the release notes for 2.4. IIRC, there is something to do if you're using sharding or it will break, but replica set are fine. keep in mind it is better to migrate all nodes ASAP, partially migrated state is not meant to last long
[09:20:03] <Seraf> hello everyone. I have a replicaset. All nodes are stuck in STARTUP2 mode ... I know which one should be the primary, how can I force it please ?
[09:44:47] <liviud> also we seem to encounter a problem with a custom php app that's using the 1.4.3 driver
[09:45:31] <liviud> in the replicaset environment, if one of the nodes times out ( for example by lack of connectivity ) the driver seems to get stuck in the connected state for a large while...
[09:46:37] <liviud> which, although we used the default timeouts code-wise, it should be by default less than half a day I suppose :)
[09:47:20] <liviud> on the other hand if the node goes down ( kill -9 mongodb or stop it ) everything runs just fine
[14:54:32] <Industrial> Hi. I have a user collection for my webapp and I have a complete authentication setup now, however the authorization is limited to 'are you logged in or not'. How do I authorize users to perform requests/actions in my webapp based on a structure/collection in mongodb? Anyone done this before?
[14:55:14] <Industrial> I want to be able to specify that users can view all profiles but only edit their own profile, and a moderator or admin could do everything
[15:10:57] <Nodex> Industrial : I do that with an array of users / groups and interset the current user to it
[15:18:21] <edugonch> Hello, I'm having some problems with rails 4 and mongoid, I'm getting this errors http://pastebin.com/8RaaSMDH
[16:15:37] <sflint> question...dropDatabase gives space back to the OS...but remove() on a collections does not give space back...right?
[16:38:00] <adamcom> if you removed everything then did a db.repairDatabase() that would reclaim the space too, but the drop is going to be a lot quicker
[17:07:19] <tiller> oh shi-. I'm sorry cheeser! I used this method a lot for updated, never seen the remove flag =/
[17:35:14] <insanidade> hi all. I'm using a script to restart and connect to mongodb but I have the following issue: the service restart output reports that mongodb was correctly restarted but any attempt to connect to it results in an output as if mongodb is down.
[17:36:48] <insanidade> tiller: shall I use some sort of "sleep" command? how long would you suggest?
[17:38:06] <insanidade> Nodex: that's my script output. it works perfectly if I issue the same commands 'by hand'
[17:38:35] <insanidade> Nodex: yes, it does. I'm pretty sure due to other commands that need root permission being correctly executed.
[17:40:22] <insanidade> Nodex: I just tried something to check if I should add some sort of 'wait' before mongodb is up. I think that's the case.
[18:12:58] <quickdry21> How can I convert a replica set to a standalone in the config DB? I'm trying to test backups but don't want to have to deploy an entire replica set for each shard.
[18:28:51] <quickdry21> To answer my own question, yes, you can do a config.shards.update({_id: 'rs*': 'new-host:port'}) for each shard to convert it to a standalone and change the hostname all at once, then copy the configdb path to all three config servers
[18:54:35] <a|3x> what could be wrong with my text search attempts? http://pastebin.com/GHkDbfP9 why is it not finding my document in both cases?
[19:07:13] <a|3x> cheeser, "default language of the index" sounds like an index property, not collection document field
[19:09:02] <a|3x> where does it say in the docs you can have documents with different language settings?
[19:09:07] <cheeser> http://docs.mongodb.org/manual/tutorial/specify-language-for-text-index/#specify-the-index-language-within-the-document
[19:13:24] <a|3x> also, another problem is what i have actually, i am using 'language' for something else, i was debugging a problem with text search for hours until i found that my own language field is breaking things
[19:15:09] <a|3x> if you are going to do this, maybe allow db.quotes.ensureIndex( { quote: "text", language_field: 'blah' } )
[19:27:01] <cirwin> Is there any reason to use config passed in as a hash instead of in the URL? I'm using the mongoskin node driver to connect to a mongos and want to specify readPreference=secondaryPreferred
[19:27:17] <cirwin> it seems like just putting in the URL shoudl work, but I'm not sure how to test it
[19:52:09] <bjori> qswz: for example, if the user input is used to structure your fieldnames and criterias then you are in big troubles
[19:52:46] <bjori> same with values. if the user changes the input to an array, then he can create a complex query that does totally different things then you wanted
[19:53:20] <bjori> so if you are doing fieldname $eq user-input, you better be suer user-input is a scalar value, not array
[19:55:16] <qswz> I'm mapping ajax request in json, more or less directly to mongo's requests, even the collection [collectionName, 'find' or 'update'.., json, orther json]
[19:58:08] <qswz> then there is access right on documents, with hard-coded _canRead, _canUpdate fields, to restrict documents to a group
[20:01:57] <bjori> please don't use a user defined variable for picking operation or collection name
[20:13:42] <qswz> If I create a json with a valid token, I'm able to put _canUpert in the document coainting my email and others, and only those can upsert it, same goes for delete
[21:53:10] <travisgriggs> i'm changing my document schema, changing to custom _id's. Is there a way to find() all docs that have an _id of type ObjectID? Or another technique to remove all of them?
[22:27:01] <travisgriggs> i just learned that remove() actually does a match, you don't just have to feed it one complete document at a time. i seriously love mongo
[22:44:18] <travisgriggs> having just switched one of my doc fields to be the _id... I find myself missing the more descriptive name. I wish there was a way at collection creation time to indicate an alternate name for the sacred/special _id field
[22:46:03] <jjbean> Hi, would like to offer advise about a making a geoWithin query through mongoid?
[22:46:21] <retran> i can't think of any benefit i've derived from having PK have an arbitrary name, in RDBMSes
[22:47:06] <retran> it's just one more stupid thing to familiarize myself with in a schema that has nothing to do with business logic
[22:48:01] <iwantoski> When I'm updating a document, it removes any field that isn't present in the update - why is that? Much I get the full object in order to preserv its completness
[22:48:22] <jjbean> *meant 'would anyone'. I can successfully make a geoWithin query from mongo shell, but not through mongoid
[22:49:07] <tystr> TypeError: Cannot read property 'version' of null at src/mongo/shell/utils.js:1008
[22:52:09] <Joeskyyy> iwantoski: http://docs.mongodb.org/manual/reference/method/db.collection.update/
[22:52:23] <Joeskyyy> By default it'll do that, if you wanna change just one part of the document, use $set
[22:52:27] <retran> you know that ORMs will remove some capabilities, they may not have every single feature properly abstracted
[22:52:44] <jjbean> I did not see why not to use it. I saw how I could fit the pieces of my web app togeather and decided to use it
[22:54:52] <jjbean> I think you might be right about the geoWithin query feature in my case, but thought I would check here as i'm running out of ideas at the moment
[22:54:59] <travisgriggs> consistency is nice. it seems like you just trade "just one more stupid thing to familiarize myself with in a schema that has nothing to do with business logic" from one to the other
[22:55:35] <travisgriggs> you never have to worry about figuring out what the special/sacred field is. it's always the same. otoh, any time it's custom, you have to figure out what it actually is
[22:55:45] <iwantoski> Joeskyyy: Oh wow, I missed that .. I've been look at that page for a while now :/
[22:56:02] <travisgriggs> any code you write around it, you have to figure out "so what are they using _id for here?"
[22:56:25] <Joeskyyy> But ultimately it makes sense to update by changing the document by default, then using $set in all other cases.
[22:56:54] <travisgriggs> so i offer my second wish, slightly more educated now :) i wish there were a way to have field aliases. Then you could have _id always be there. but one could alias it where appropriate
[22:57:55] <iwantoski> Joeskyyy: Yea sure, I'm not one to critize that solution because I'm new to both mongodb and nosql/non-relational dbs in general - I'm coming from a traditional setting with relational dbs :)
[22:59:54] <iwantoski> Hm, "while I have you one the line", I have a different question I havne't had time to test yet. regarding 2d Indexes, which I've only read about - but they seem to be sub-optimal when latency is of importance, any thoughts/input regarding this?
[23:01:33] <iwantoski> travisgriggs: I'm new but _id is just a PK, you can modify that if you need AFAIK?
[23:02:30] <iwantoski> retran: The impression I've got is that it's not as fast as - say postgresql's implementation
[23:03:12] <iwantoski> (i'm going to do this using mongo for now anyway to try it out, but I just thought I could get some feedback on the matter)
[23:04:28] <iwantoski> I'm not trying to call out a db war, I'm just asking - because for what I'm building, I'm really concerned about low latency lookup from (2d) geospatial data which is high concurrent.
[23:04:34] <travisgriggs> so you see one of my documents {'_id': 'F00008', 'last_update': <datetime>, 'ip_address': '10.20.30.40'}, and you know what I'm using for my PK retran?