Log file Viewer
#mongodb logs for Tuesday the 12th of June, 2012
[02:24:20] <krz> http://pastie.org/4071109. it will always push to the array 'visits', if visits is an existing array. which is fine. but what if i wanted to $push to visits, only if name foobar did not exist. if it did, then just update the token. what should i be using?
[02:31:24] <whaley> is there a way to initiate a replica set on a database from the java driver? (this is for some system tests where I'd like to have a replica set available but created at runtime of the tests since the databases are setup/teared down after test runs)
[02:33:34] <whaley> I'm going to assume it's a DB.command("rs.initiate({configstuff})"). Going to give that a shot.
[08:39:24] <Saby> hi, i was wondering if it is possible to write a query which updates all records whose currentDate - previousDate > 2 hours in a collection
[08:43:18] <carsten> nothing to be explained. this can not be done client side except using the $where operators which is not recommended
[09:00:27] <Saby> so basically i should retrieve all the records and do the calculation in my program ?
[09:01:49] <Saby> this thread would keep running every one hour, so that is why i wanted to get it done on mongo db side
[09:05:32] <NodeX> [10:01:47] <NodeX> to put it simply .. without map/reduce Mongo will not calculate things for you7
[09:08:11] <NodeX> I have carsten on ignore because he is annoying so I don't know what advice he is offering
[09:22:14] <horseT> Hi, I need some help regarding the aggregation framework, I want to do somethink like Select sum( point) from score GROUP BY date;
[09:34:00] <Saby> how should i write the reduce function, which basically checks if previousDate-currentDate > 2 hours then set priority: 2
[09:47:42] <Saby> I'm assertion" : "invoke failed: JS Error: TypeError: value.previousDate has no properties nofile_b:2",
[09:47:50] <Saby> I'm getting that error >> assertion" : "invoke failed: JS Error: TypeError: value.previousDate has no properties nofile_b:2",
[10:12:12] <siimo> is there a sqlite style mongo version ? where i can store everything in single file and use a library to query it?
[11:00:14] <Saby> Carsten: any insight why this MR won't update the records ? http://pastie.org/4072721
[12:29:16] <spillere> i'm really new to mongodb, i have a simple question, i'm doing a foursquare app, and i want to save the user and all his checkins in a db
[12:30:57] <spillere> { user: 'name', checkin: [ name: [ 'velue1', 'venue2'], latitude : ['lat1', 'lat2'] ] } ?
[12:41:05] <spillere> is this right for a database scheme? http://pastie.org/4073261 as for one user will be lots of checkins
[12:44:01] <spillere> I wanna query the user info and ALL the checkins, doing a FOR and displaying all checkins
[13:23:25] <Industrial> Where did I go wrong with installing mongodb? I have both mongodb-server and mongodb-clients installed and the mongodb server is running but I don't see it in ps -e|grep mongo https://gist.github.com/1014dc4231f55a392f5f
[13:29:38] <multiHYP> i have a complaint to make. why is the java driver not supporting nested field access like the mongo shell?
[13:30:07] <multiHYP> java driver example (which is not working): new BasicDBObject("name.title", "some new title")
[13:36:24] <kali> multiHYP: new BasicDBObject("name", new BasicDBObject("title", "some new title")) :P
[13:37:02] <multiHYP> how many more nested BasicDBObjects would you define until you'd think enough is enough?
[13:37:44] <multiHYP> my collections look like crap because of this shortcoming <not impossibility> in the java driver. I thought it was self evident.
[13:45:04] <multiHYP> so here I read that the java driver is just a json parser: http://stackoverflow.com/questions/3580529/how-do-i-update-fields-of-documents-in-mongo-db-using-the-java-driver brilliant, what an achievement. :(
[13:46:24] <algernon> ("name.title" at insert time works the same in the shell as it does with java. that BasicDBObject(...) is longer than {...} is another detail.)
[13:49:53] <maik_> We have a thread running, but he is waiting for whatever. Last line of threadump:
[13:49:53] <maik_> java.lang.Thread.State: RUNNABLE at java.net.SocketInputStream.socketRead0(Native Method)
[13:52:09] <multiHYP> ok here i have this: {"detail": {"title":"a", "name":"b", "address":"c"}} how would you $set the title only?
[13:54:01] <multiHYP> algernon: no way, i don't know whether you used the java driver at all. java driver does not even support "detail.title" type of notation.
[14:00:58] <multiHYP> I put my case back on the table. here is something perhaps new to algernon too
[14:02:16] <multiHYP> strictly java driver related: if you use $set, the "detail.title" notation nested in a "$set" works. otherwise it doesn't. So the only alternative in those cases is to pass a json object very similar to the shell command.
[14:03:28] <algernon> multiHYP: I still don't see your issue. update expects dotted notation, insert doesn't. same is true for the shell.
[14:04:23] <multiHYP> upsert without $set in shell supports dotted notation, but in java driver doesn't. <- I cannot express it more clearly than that.
[14:05:47] <spillere> im trying to connect using flask/python using db_connection = pymongo.Connection()
[14:09:22] <Ratler> Could be that a leading number just doesn't work using attribute style (not a python expert) :)
[15:17:49] <m_rec> hi everyone, is there a reason rs.status() would no longer return stateStr when being run? since the replica set elected a new primary I can't seem to find that field anymore
[16:45:20] <jasiek> if i have a data migration that removes old versions of documents and inserts new ones, am i likely to benefit from batch inserts?
[16:59:37] <c3l> If I have, in my collection, a document like the following, how do I uniquely identify a "subdoc"? {title:"Some Title", subdocs:[ {title:"Subtitle one", title:"Subtitle two"} ]} For instance, there is nothing stopping the subdocs from having the same title.
[17:02:08] <c3l> NodeX: Im quite new to mongodb, I dont really understand. Yes I would like to have some unique index for each subdoc. Should I manually set a "_id":ObjectId(..) key/val pair for each "subdoc"?
[17:05:38] <c3l> I cannot ensure that the titles (or any other field for that matter) will differ, since these can be manipulated by the user. What would be the best way to ensure a unique index on it?
[17:43:24] <jshanks> I found that there is a hex_md5() function available on the server (for use in server side javascript), is there also a function for sha1 built in?
[17:44:45] <dotty> Hi everyone. I want to ask a question about MongoJS (Node.JS library) - should I ask here or in #Node.js?
[17:48:48] <dotty> I'll ask in both I guess. I'm attempting to connect to MongoDB using the following: "require('mongojs').connect('localhost');". This worked on my work machine but, now I'm on my laptop, it's not working and I'm getting the following error: "database name cannot be the empty string" - any clues as to why?
[18:26:05] <kali> jshanks: any javascript implementation should work, but i don't think there is anything builtin
[18:27:19] <jshanks> ok thanks kali. since there is a built in "hex_md5" function, I was hoping there was a built in sha1 function as well.
[18:29:29] <jstout24> question… which schema is better… we have visitors with attributes… here: https://gist.github.com/e185cc4a565948b00937
[18:34:22] <kali> jstout24: I don't like to twist the model... I wouls also use a boolean there instead of a string
[18:35:21] <jstout24> kali, yeah, currently it's up to the user to create "checkboxes" or form fields and those values are stored
[18:35:48] <jstout24> but we can def do checks on the application layer to do: if ('yes' === $value) $value = true, etc
[18:40:19] <jstout24> i wonder the performance of `db.visitors.find({ email: 'some@email.com', 'attributes.likes_fat_chicks': 'yes' })` vs `db.visitors.find({ email: 'some@email.com', 'attributes.likes_fat_chicks': { $in: ['yes'] } })`
[18:42:36] <jstout24> nice… so my application layer doesn't have to figure out which to use for a performance gain :)
[18:43:21] <jstout24> kali, i just realized i used my name in all the examples and i admitted to liking fatchicks
[18:46:29] <mkmkmk> i have a collection that has 1000-2000 writes/sec… they're actually upserts.. think of it like a unique visitor log.. if the entry doesn't exist, it's created and the _id is a standard mongo object id.. if it does, i $push a small string onto an array on the document.. im working on sharding it, but i need to figure out a good shard key for the visitors collection
[18:46:56] <mkmkmk> since im using objectid's for _id, that falls into the trap of pushing most of the writes to a single shard
[19:02:46] <dstorrs> another option would be to include an arbitrary field to the record that comes out of a call to rand()
[19:03:40] <dstorrs> fwiw, I'm in the same boat as you -- just migrating to Mongo, haven't done sharding yet, still figuring it out.
[19:04:38] <dstorrs> and you must be storing SOMETHING other than the _id, or there's not much point in the collection. what else is in there?
[19:48:56] <dmuino> i need to do a mongorestore of a large collection on a sharded cluster. The collection + index will not fit on a single replset. I'm currently doing the restore but it seems that all the inserts are going to the first shard, and that will fail
[19:49:23] <dmuino> is there a way for me to get the writes going to all shards? (using mongo 2.0.5)
[19:54:33] <cjm> I'm getting this error, ReferenceError: message is not defined when trying to do a geonear command
[19:55:18] <dstorrs> dmuino: It may well be that your shard key is such that all writes default to the right-hand shard.
[19:55:43] <dstorrs> In which case, the only solution (AFAIK) is to let the balancer shift chunks around to make space.
[19:57:28] <dstorrs> dmuino: yes, but remember how the sharding works -- mongos looks at the shard key and writes the data to a chunk on the appropriate shard. when a shard starts to fill up, the balancer moves chunks to other shards.
[19:58:28] <dstorrs> if your shard key is something that constantly increases (e.g., a date), then all data will initially go to the right-hand shard
[19:59:16] <dmuino> it's a compound key - i have 'name', 'cluster', 'country' - and the data should be quite random
[20:00:11] <dstorrs> http://www.mongodb.org/display/DOCS/Choosing+a+Shard+Key#ChoosingaShardKey-Writescaling
[20:47:29] <Xxaxx> hello. how can it be, after db.copyDatabase('vk_photolike', 'vk_photolike', '188.93.20.75') , source: vk_photolike 65.921875GB, destination: vk_photolike 7.9501953125GB
[20:55:51] <leandroa> hi, I have a doc like this: https://gist.github.com/2920068 and I need to add indexes for sorting by sortable_key.en-US and sortable_key.es-AR. What is better, an index to 'sortable_key', or many indexes by keys inside sortable_key?
[21:08:35] <BobFunk> having a really simple query on a pretty small capped collection take incredibly long, even though it should be a simple index lookup
[21:14:42] <BobFunk> and when I run the query from the console, its fast enough and seems like its using the index
[21:21:25] <jstout24> Q: if i'm storing data as shown in method 2 (https://gist.github.com/e185cc4a565948b00937), how do i index attributes properly?
[21:21:46] <mediocretes> I'm pretty positive the problem is that it's not using the index, and I don't know why. Good luck!
[21:22:05] <jstout24> for a query like this: `db.visitors.find({ 'attributes.favorite_foods': { $in: ['pizza', 'beer'] } })`;
[21:22:38] <mediocretes> jstout24: http://www.mongodb.org/display/DOCS/Indexes#Indexes-IndexingonEmbeddedFields%28%22DotNotation%22%29
[21:37:42] <BobFunk> hmm, ok - it seems that when I pick a new url the query takes forever, when I pick one that has been queried before, mongo uses the index and it's fast enough
[22:01:52] <jstout24> i need a mongodb genius… here's my problem and attempts at it: https://gist.github.com/eb72b04ce8990bd35b7e
[22:28:55] <BobFunk> even a query with a hinted index still takes forever - small collection, indexed field, straight forward query - 56 seconds
[23:17:13] <grafman> Mongodb newbie: I want to iterate over the all of the collections in a db. I'm getting the names using x = db.getCollectionNames(). I'm not seeing any examples on how the use the results or of a better way to interate over all of the collections in a single function.
[23:31:32] <grafman> I'm a mongodb beginner. I want to iterate over the array db.getCollectionNames() gives me. Does anyone have a pointer to how this is done?
[23:32:00] <mediocretes> it's javascript, all the usual array methods apply, because it's javascript
[23:33:27] <mediocretes> note: there is nothing special about forEach or the function being passed to it. l = function(x){ … } ; db.getCollectionNames().forEach(l); will work just fine, because it's just javascript
[23:34:28] <mediocretes> you might also enjoy this: for each(var x in db.getCollectionNames()){ print(x);}
[23:35:36] <grafman> Is there a way to iterate over all of the collections? I have a find that will return the last inserted document that I want to run on each one.
[23:37:53] <mediocretes> there's also this, saved me a bunch of time: db.getCollectionNames().filter(function(x){return x.indexOf("A") == 0})
[23:38:26] <mediocretes> in the shell? write a function that does what you want and takes the name of a collection, then pass it to foreach
[23:39:06] <grafman> db.getCollectionNames().forEach(function(x){ db.getCollection(x).find().sort({_id:-1});})
[23:39:36] <mediocretes> so, you aren't outputting the results, and those results will be sort of ugly
[23:41:30] <mediocretes> also, are you sure _id is in order? do you have a timestamp field or something?
[23:48:18] <grafman> Ok, I asked the question there and I'll credit you with the answer if you know the account name you use there
[23:51:50] <grafman> my manager walked up to me and said we need a report of xyz from mongo, that one Bill used to do