Log file Viewer
#mongodb logs for Wednesday the 7th of November, 2012
[00:15:33] <dblado> in the product category use case would you add a 'root' category so even a top level category would have a 'parent'?
[00:17:24] <FrenkyNet> Derick: I've found a bug in the php RC, the connect option is either ignored, or the reporting is wrong: http://snipr.it/~qr
[00:26:11] <astropirate> Can the size and count of a Capped Collection be changed after it has been created?
[00:26:21] <FrenkyNet> for who's interested, I've created an issue for it: https://jira.mongodb.org/browse/PHP-560
[00:39:56] <tjr9898> I'm working on dev homework 2.2 and am able to print out the sorted hw scores with pymongo. I can't get the update to work by using the same iterate
[01:03:49] <bjori> $ php5.4 -n -c php.ini -r '$m = new Mongo("standalone", array("connect" => false)); var_dump($m->connected);'
[01:06:38] <bjori> ckd: do you mind running MongoLog::setLevel(MongoLog::ALL); MongoLog::setModule(MongoLog::all); $m = new Mongo("..", array("connect" => false)); echo "Dumping prop\n"; var_dump($m->connected); for me and pastebin the output?
[01:21:14] <bjori> $ php5.4 -n -c php.ini -r '$m = new Mongo("primary", array("replicaSet" => true, "connect" => false)); var_dump($m->connected);'
[01:28:52] <bjori> but if you are running an app on the same server, and hit a process that already had an connected mongodb server it will still be connected
[02:42:41] <zastern> I'm trying to set a variable in the mongo shell equal to the contents of an object returned with a db.foo.find({ some_condition }). I'm doing it like this but it's not working, any thoughts? t = db.products.find({_id : ObjectId("507d95d5719dbef170f15c00")})
[02:51:19] <mrobben> mongoose is ODM based - you create a schema for the collection you're trying to call 'find' on.
[02:51:38] <mrobben> then you instantiate the schema to create a model and apply the operation to that.
[02:52:03] <mrobben> if you have an existing schema defined, throw it in a gist and I'll take a look
[02:52:09] <ttdevelop> what i don't really get from examples is there is little mention on collections
[02:52:26] <ttdevelop> schema looks like it is the properties for the collection, like name, address
[02:54:20] <mrobben> when you create the schema instance, it automagically maps that to a collection for you.
[02:55:28] <ttdevelop> that's what i'm trying to do from what i learned here : http://stackoverflow.com/questions/5794834/how-to-access-a-preexisting-collection-with-mongoose
[06:35:16] <ryn1> guys,. pls help, im trying to export data from mongodb, using this query --> mongoexport -d dbName -c myCollection -f title,meta,contents --csv -o test_explort.csv,. my problem is, the content field, in the exported file doesnt get the complete data content and convert the other part of the content to ellipsis..
[06:39:58] <crudson> what types of objects are meta and contents? (j|b)son doesn't translate to csv particularly well except for the simplest of data types
[06:42:33] <crudson> have to run, but I bet that it's some nested attribute, so I'd advise using json as the export format
[08:37:10] <samurai2> hi there, is there a way to change some field value in input collection while we're still in the map phase of the map/reduce? thanks :)
[12:10:18] <Danielpk> How i can start mongodb at OSX? i tried "mongod --fork --logpath /var/log/mongodb.log --logappend" but don't started. Any tip?
[12:11:40] <Danielpk> Humm i got this error "exception in initAndListen: 10296 dbpath (/data/db/) does not exist…."
[14:21:12] <eka> hi all... I see that .hint() on a find improves the performance of my query. Is there a way to give hint() to aggregation or map/reduce?
[14:30:02] <Derick> the connect option isn't part of the "connection string" but it's still interpreted later
[14:38:10] <ron> Derick: ORA-600 is oracle's way of saying 'Something went wrong, it's really bad, but we have no idea what the problem is, so you're pretty much screwed. Good luck!'.
[15:42:59] <sander__> Do anyone know about a good beginners guide for mongodb and nosql.. for those who knows SQL good already?
[16:01:45] <squawknull> sander__: 7 databases in 7 weeks is awesome... http://pragprog.com/book/rwdata/seven-databases-in-seven-weeks
[16:01:59] <squawknull> it's not specifically on mongo, but it's a good primer on nosql in general, and the different approaches
[16:03:16] <squawknull> sander__: fowler's book is also supposed to be really good. i've not gotten around to reading it yet. http://martinfowler.com/books/nosql.html
[16:39:33] <prune> amybe one of you will help me build a good schema for a foursquare like application
[16:40:54] <prune> as people do "checkins" but never "checkout", I'm wondering where to keep the checkins (which is a triple of user, location and date) and where to keep the current user count on each location (which is equal to the number of checkins in the last 30 minutes)
[16:41:58] <prune> For the moment my idea is a collection for places, a collection for users and a 'live and highly changing' collection for checkins, where I may duplicate some informations
[16:43:39] <prune> eka : but how will I display the "checkins count" for every places ? a map/reduce every 5 minutes in a backend task ?
[16:44:44] <prune> I need the number of checkins in the last 30 mins for each place to display on the map
[16:46:06] <ckd> I was just making sure I can duplicate the orig problem on a different box, and it reports i'm running RC2-dev
[16:53:30] <ckd> Derick: Yep, you're right… I just never noticed it, RC1 does report itself as RC2-dev
[16:59:01] <JakePee> Does anyone have any insight regarding the impacts of highly volatile collections, essentially used as a cache/staging area
[16:59:12] <FrenkyNet> Derick: what are the default levels for MongoLog? I removed the log setting but the settings remain the same lol
[16:59:39] <JakePee> Collections would be filled every 10 seconds or so by a threaded application, then processed and dumped every 30 seconds
[17:11:02] <podman> I've got a problem that I'm trying to track down and I'm really not sure where to start looking. I have a ruby processes that pops messages off of a queue and then upserts documents in mongodb. I have this process running on two different machines. one runs fine, but the other one occasionally gets stuck for about 10 seconds while updating a document.
[17:19:55] <Derick> ckd: make sure you have all your replicaset members in your connection string for now
[17:22:10] <podman> a few updates are taking a little more than 100ms, but the client is hanging for 10 seconds
[17:24:19] <podman> i would have thought it was a locking issus, but you'd think both clients would be affected. it seems to only be one client though
[17:24:39] <ckd> yeah, it's not super mission critical for me, I was just using that second connection to write out logs to a separate instance, but it was on my list to take that out of the path anyway
[17:32:10] <ckd> that test code, I wrapped it in a for block to insert a record 1000 times… mongo keeps trying to use the same id, so it's failing
[17:48:36] <ckd> oh, interesting, I never realized that the driver adds the _id to the original object
[18:41:18] <AugCampos> I there is possible to get all diferent values for a field in all docs in a collection?
[18:43:49] <AugCampos> ex: { name: "zzz", type: ["typea", "typeb"]} , { name: "zzz2", type: ["typea", "typec", "typed"]} and retrieve ["typea", "typeb", "typec", "typed"]?
[18:46:19] <AugCampos> I there is possible to get all different values for a field in all docs in a collection? ex: { name: "zzz", type: ["typea", "typeb"]} , { name: "zzz2", type: ["typea", "typec", "typed"]} and retrieve ["typea", "typeb", "typec", "typed"]?
[18:46:49] <choover> Hello! Noob here... I am trying to set up a replica set on my mac. I want all three mongod servers on my localhost to be part of a replicaset per these directions here: http://docs.mongodb.org/manual/tutorial/deploy-replica-set/ ... I have followed the instructions succesfully up to step 6 where it says to add members to the replica set you need to call the add method on the rs object like: rs.add("localhost:27018") ... When I at
[18:46:54] <choover> can't use localhost in repl set member names except when using it for all members
[18:53:45] <psyprax> I'm interested in storing a large (perhaps 1M by 1M) 2-dimensional matrix (not sparse, but not completely dense either) whose cells will be updated frequently
[18:55:49] <kali> psyprax: if you store it in forms of a collection of (x,y,value), you'll have a big overhead
[18:55:53] <psyprax> kali: actually it would need to be read frequently too since I'm interested in updates, not just writes
[18:57:15] <kali> psyprax: maybe you store it per row or column, depending how dense it is. each document in mongodb must be below 16M
[18:57:46] <kali> my advice if you can choose the techno, don't make this your first experience in mongodb, as i fear it will be painfull :)
[18:58:53] <psyprax> kali: haha, ok. Thanks! I'll experiment a bit with an (x,y, value) representation, but will probably go with something else as you said
[19:18:00] <AugCampos> I there is possible to get all different values for a field in all docs in a collection? ex: { name: "zzz", type: ["typea", "typeb"]} , { name: "zzz2", type: ["typea", "typec", "typed"]} and retrieve ["typea", "typeb", "typec", "typed"]?
[19:24:11] <AugCampos> crudson: in the examples in (http://www.mongodb.org/display/DOCS/Aggregation#Aggregation-Distinct) only show for value fields for arrays works too?
[19:30:39] <tystr> info DFM::findAll(): extent 4:5c117000 was empty, skipping ahead. ns:local.replset.minvalid
[19:32:03] <crudson> AugCampos: well if you're too busy then I did try it, and here is the result: http://pastie.org/5342020
[19:34:04] <tystr> well our traffic is a bit higher today, and this is the first time I've seen this error
[19:37:36] <ckd> well, if you're seeing high traffic today and def not doing deletes, then slow queries seems like the obvious culprit
[19:38:09] <ckd> i don't have the luxury of high traffic yet, so I haven't encountered that, but I'd be curious to know what caused it for you in case i ever do see it
[19:40:00] <crudson> AugCampos: I didn't necessarily expect it to work, but always worth trying. The docs are good, but not always complete.
[19:45:37] <podman> Derick: issues with a geoip gem i'm using. for some reason some lookups are taking almost 10 seconds
[19:51:18] <podman> going to try launching an instance with ephemeral storage and see if that does anything to help
[20:34:06] <JakePee> anyone know if it's better to allow php's built in connection pooling manage multiple connections or to just use
[20:54:43] <ckd> JakePee: They, among other things, rewrote the connection stuff… Hannes is doing a talk on it next week: http://www.10gen.com/events/webinar/whats-new-php-driver
[21:02:08] <JakePee> ckd: I actually just threw together some tests against it with 100,000 iterations doing a findOne against the same collection. If i create a connection on each iteration, it took 7.68110585213 seconds. If I reuse the same connection, 6.49627995491 seconds.
[21:02:36] <JakePee> not exactly an exhaustive test, but show that there's not a lot of overheard on creating a connection
[21:04:19] <JakePee> also, if you don't dereference the connections, a single pool only handles up to ~190 connections
[21:05:33] <ckd> "The new framework no longer has the concept of a connection pool, but instead make sure there is only one connect per node/db/username."
[21:08:37] <ckd> JakePee: hopefully that'll save you some time, the issue is somewhat academic at this point :)
[21:27:55] <JakePee> ckd: thanks, I was more curious if there was any reason to leverage the connection pooling but it would seem to be more of a beneficial exercise for a threaded environment
[21:45:38] <ckd> well, let's just say that I tested beta 1 and 2 for about 5 minutes each before I deployed 'em in production
[21:49:24] <ckd> although that cluster is pretty much doing all reads, so I figured nothing too horrible could happen
[21:51:56] <ckd> speaking of which… so tossing just one host from the replica set doesn't have any issues for me
[21:52:33] <ckd> but passing a string like 'mongodb://host1:port,host2:port,etc' to the constructor makes things cranky
[21:59:41] <ckd> i vaguely remember you suggesting something about it only working if i specify all members
[22:44:06] <rshade98> db.command({ "replSetInitiates" : { _id: 'repl1', members: [{_id: 0, host: '10.202.18.48'},{_id: 1, host: '10.144.153.251'},{_id: 2, host: '10.144.15.96'}, ] } })
[22:45:12] <cjhanks> Hey all -- with the C++ version client drivers; is it possible to set the FD_CLOEXEC flag on the DBClientConnection with out modifying the driver source code?
[22:52:14] <cjhanks> In fact I can't find the tcp code in the source. Any suggestions would be useful.
[22:54:21] <Guest19926> Installed MongoDB on a 32-bit debian machine via 10-gen package. When I start mongo, it drops after about 30 seconds without warning. Before it does, I am able to visit the web admin and see things. Any ideas as to why? Thanks.
[22:55:12] <cjhanks> Guest19926: My experience is the packages are terrible and you're better off just building it yourself.
[22:56:10] <Guest19926> cjhanks: Thank you for that knowledge. I will definately try self installation.
[23:15:53] <Guest19926> ckd: Lest couple messages say: connection accepted from 192.168.2.159 (My lap top while on the network) Invalid operation at address: 0x819b503 from thread: conn1
[23:17:54] <ckd> Guest19926: See if this is relevant to your specific setup: https://jira.mongodb.org/browse/SERVER-7012
[23:18:07] <Guest19926> ckd: ALso not, I do get a warning on startup: 32-bit servers don't have journaeling by default. PLease use --journal if you want durability.
[23:21:49] <ckd> Guest19926: That processor is from '03, I think it's reasonable to assume you could be running into that same issue as in the reported issue
[23:24:26] <ckd> Guest19926: I looked at some of the related issues, and it would seem that your processor isn't supported
[23:25:04] <ckd> Guest19926: There's an open ticket to have Mongo check if it's running on an unsupported CPU, but it's not implemented yet
[23:37:33] <cjhanks> ckd: That is, 10gen RPM for Fedora hung the init system and was completely screwed. The Debian package require reconfiguring, though it didn't hang the init.
[23:45:57] <spartango> hi, new to mongo (coming from cassandra)...question about data model: if i have a list of strings that i want to associate with an id... is storing them as keys in a document with no/empty values looked down upon?
[23:54:03] <spartango> its not all that different, i could definitely do that. is there a particular reason to store them in an array?
[23:55:53] <rossdm> well you said that you have a list of strings, which makes me think that they're related in some way and it would be nice to iterate over them. Making them keys in an object with null values sounds like it doesn't represent the data properly. Hard to say more without knowing specifics
[23:57:41] <spartango> they're URLs for parts of a file, accumulated over time. so you can imagine that i'll want to append to that list a bunch of times, but when the parts are their i'll probably grab all the strings at once