Log file Viewer
#mongodb logs for Thursday the 30th of May, 2013
[00:01:31] <foxer_> Hello guys, I boast of Ramnode, my vps makes 1.3GB per second. For quite some time I am their customer and for me this is the best hosting company! I'll be glad if someone signs up from my referral link and become a happy customer of Ramnode! https://clientarea.ramnode.com/aff.php?aff=414 So, performance for mongodb is extremely important!
[00:01:59] <scottbessler> does a rs node that is in RECOVERING accept writes? (e.g. would wc majority fail if i have 1 primary and 2 recovering) ?
[01:02:15] <sinusss> hi. will I have issues if I delete all (children), then re-adding them instead of updating each child?
[01:34:45] <sinusss> works nicely... I can CTRL + F on items and it feels like phpmyadmin (i like phpmyadmin)
[03:37:36] <frozenlock> Hello gentlemen! I want to configure one of my machines to run a replica set. However, mongod starts automatically when Ubuntu boots (without the --replSet argument). Do I need to remove the auto-start, or is there a way to change the default configuration?
[04:55:59] <frozenlock> Hmmm... my empty secondary doesn't seem to fetch the data from the primary... or at least not very fast (I'm looking at my network usage). How can I check if the secondary is indeed catching up?
[05:02:01] <skram> Have a question. I have a master-slave setup, If I include a load balancer (which i have, just disabled). If the traffic and requests is sent to the secondary it wont accept writes, correct? How would I handle this?
[05:04:50] <belak51> I'm just getting into MongoDB and I've been playing around with it… what's the best way to store files and have a reference to them in a record?
[05:04:55] <belak51> I know I can use GridFS to store them, but how is that commonly stored with other data?
[05:51:11] <badshaah> Hey, I am trying to start mongodb in clustered mode with replicaSet, on openvz
[06:42:37] <belak51> Does a GridFS ContentType have to be a valid mime, like text/xml? Or application/pdf? Or can it just be something like the extension?
[07:36:38] <resting> so i'm looking at the example of mapreduce http://docs.mongodb.org/manual/reference/command/mapReduce/#calculate-order-and-total-quantity-with-average-quantity-per-item
[07:40:18] <crudson> resting: examples are in javascript, mongodb's scripting language, so they can be entered into the shell 'mongo'. If using a different language consult that language's client API docs for syntax.
[07:43:27] <resting> crudson: i see…is there anyway to store these javascript commands as some sort of script file and have our application execute it instead?
[07:48:12] <resting> Nodex: thanks…so much stuff…think i'm losing my ability to read…*continues to stare at screen
[07:49:38] <crudson> resting: you can put in a file and do: load('somefile.js') from the console, that will make the functions available in the shell
[07:50:29] <crudson> makes multi-line editing easier rather than 'edit function' in shell which is volatile and discards unparsable edits
[07:51:22] <Nodex> resting : yes, yu build a query and it returns your result, it's much faster than map/reduce
[07:51:35] <crudson> if your "application" wants to use them, then that depends on your architecture - http://docs.mongodb.org/manual/core/server-side-javascript/ may help you
[07:52:57] <crudson> but as nodex implies - the problem you are trying to solve may determine whether to use m/r or aggregate.
[08:00:16] <crudson> resting: yes, when you use $group http://docs.mongodb.org/manual/reference/aggregation/sum/#grp._S_sum
[08:03:42] <r0goyal> I am trying to set up mongodb replica set on openVZ machines. Mongo instances are working fine in single mode but as soon as I setup them for replica mode, it fails to start. What could be the problem ?
[08:06:04] <Nodex> there has to be somehting in the log, mongo doesn't just error without some verbosity
[08:21:10] <resting> hm..so i'm trying to sum a field within the document which is a huge object…packets: { "0": "0", "1": "200", "2": "133" ….. }
[08:21:20] <resting> db.mb.aggregate({ $match: { capture : "1369885967" } }, {$group: { _id:null, sum: {$sum:"packets"}}});
[08:23:29] <belak51> With mongo for web apps, how are the records usually retrieved? With SQL, there's usually a unique int id, but is it good practice to expose the _id of a mongo object?
[08:43:58] <resting> capture and packets are at the same level…however packets has 1 additional level with 2000 key-values
[08:46:20] <Nodex> if "packets" is a sub document then you either need to unwind it or use dot notation to reach into the object and grab the value you wish to cum
[08:47:00] <Nodex> http://docs.mongodb.org/manual/tutorial/aggregation-examples/#states-with-populations-over-10-million <--- good example to follow
[08:50:21] <resting> Nodex: i dont think its a sub document..it is a key, "packets" with multiple key-values..is that considered sub document?
[09:00:42] <Mr_O> i just imported a large har / json file in a 1 record collection. I'd like to dive into this record structure to export inner arrays into proper collections.
[09:03:41] <Mr_O> i'm new to mongodb and i need a starting point. can i expose the inner structure of a record ?
[09:13:31] <resting> Mr_O: i'm not sure if i understood "inner structure" try findOne in the shell in that case..
[09:19:11] <Mr_O> resting: i imported into mongodb a large json file into one big record. finding it makes not sense as there is only one record. i need to dive into its sub-doc hierarchy to be able to extract some of those sub-docs.
[09:22:58] <Mr_O> resting: using mysql i could ask the table description, using mongdb, how can i see a record description ?
[09:25:59] <Nodex> resting : you need to $unwind your sub document or "reach" in to it with dot notation
[09:26:29] <Nodex> Mr_O : your application can get all keys from a document - it's a trivial thing to do
[09:28:34] <resting> Nodex: :)…understand…i'd posted to SO too…http://stackoverflow.com/questions/16832035/mongodb-is-it-possible-to-aggregate-an-object
[09:29:30] <resting> lets say i could reach in, but that targets only 1 key…but i need to sum all the keys
[09:29:55] <Mr_O> Nodex: that is the issue. I don't know exactly what the doc looks like. i hope mongo would offer a "describe table" like function
[09:33:02] <Nodex> you need to adapt your data, it's also a waste of space to have numeric keys of 0,1,2,3.... etc in an object when an array will suufic
[09:39:24] <resting> Nodex: hm…i see…can i say array is easier to work with? at least in my case thats what i feel..
[10:14:21] <Nodex> for example I want to do db.foo.find({$or:[{_id:ObjectId("....")},{username:"foobarbaz"}]});
[10:17:10] <Nodex> Derick : is there a way to validate a MongoId in the php driver, for example I would like to feed it a string and have it tell me if it is or isn't a valid ObjectId before I send it to the server
[10:26:11] <Nodex> I have a query that can take an _id or a string... for example lookup by _id or by username
[10:26:50] <Nodex> and doing an $or will throw the error and is not efficient so... I would like to test the variable and do the query one way if it's an _id or the other way if it's a normal string i/e username
[10:30:04] <Nodex> I did try to getTimestamp() from new MongoId('abcd'); but it returns the current timestamp
[10:30:26] <Nodex> as this would be another way to test - i/e if the timestamp is in the past then it MUST be a valid _id
[10:31:34] <Derick> Fatal error: Uncaught exception 'MongoException' with message 'Invalid object ID' in Command line code on line 1
[10:32:19] <mtsr> Hi, I'm trying out the new text search feature, but I'm having issues with stemming. Even with language: 'none' it still seems to be doing some stemming, as searching for 'tests' doesn't work, while 'test' does, even though the text contains 'tests'
[10:33:06] <Nodex> ok so that's another way I can test it, catch the error and set the criteria in the catch() part
[10:49:37] <Mr_O> i ve got this one big document containing one field which is an array. i'd like to create a new collection made of this array elements.
[14:14:37] <marcqualie> Hey guys, what is the best way to check if config servers are in sync on a sharded cluster?
[14:20:06] <gremp> Hello, is there a way to create a function that queries a collection and modify the results before returning them ( without updating the records )?
[14:22:19] <BadDesign> How do I force mongoimport interpret a column value as string even though its a number?
[14:22:58] <BadDesign> i.e. I want to import a CSV file with the first column values having values as numbers... i.e. (line1) 2132 (line2) 32435
[14:24:17] <BadDesign> but I want to store them in the document as "value": "2132" not "value": 2132
[14:28:28] <BadDesign> BurtyB: when I get the fields value maybe I can determine if its a string or number and cast it
[14:31:04] <BurtyB> BadDesign, as I said I don't think you can with mongoimport (when I was importing data I just wrote a script to insert it with the correct type)
[14:48:48] <diegows> collection.update(dict(mail_addr='test@example.com', enabled=True), dict(status="green"), upsert=True)
[14:49:26] <diegows> db.test_upsert.update({ mail_addr : "test@example.com", enabled : true}, { $set : { status : "diego" } } { upsert: true })
[15:13:41] <diegows> collection.update(dict(mail_addr='test@example.com', enabled=True), { "$set : { "status"="green" }, upsert=True)
[15:14:47] <fxhp> dict(mail_addr='test@example.com', enabled=True) == {'mail_addr':'test@example.com', 'enabled':True}
[15:27:51] <revoohc> When failing over a replicaSet (2.2.x), what happens to writes that are issues to the cluster during the time the cluster is electing a new master?
[16:32:23] <hjrnunes> Hi all! Is there any particular reason why cursor.count() should result in AutoReconnect: timed out while next() works fine for the same query?
[17:26:03] <harenson> olegz: maybe in 7 hours when I'll be in my home. Sorry but I'm working and here I don't have data for and mongod server running
[17:30:53] <harenson> olegz: here is the database https://education.10gen.com/static/m101p-april-2013/handouts/enron.0708d010cd81.zip
[17:32:15] <olegz> harenson, probably, I can not remember, cause I pass M101 and M102 at the same time
[17:49:53] <astropriate> is it possible to pass values to the reduce function of map-reduce? i want to only get documents that meet a certain criteria. I am also referencing to to other documents using the _id field. is there some way to populate/join the document in the parent document and then run the map reduce?
[18:08:05] <daveluke> if i don't specify a field in the $set part of an update, does that field/value get removed?
[18:14:39] <belak51> What's the common way to store a reference to a GridFS file in a record such that the filename and extension can be retrieved from the file
[18:56:08] <jpfarias> so I was wondering if it is possible to have it preload all of it to make all queries instant
[19:21:53] <tystr> I added a TTL index to an existing collection , and it seems to have dropped all of the exting data
[19:23:14] <tystr> the index had an expireAfterSeconds value of "604800", and was on a "createdAt" field containing ISODates
[19:39:11] <merpnderp> if you run mongodump with --dbpath will mongod crash, or just wait for the file lock to end?
[20:54:12] <kurtis_> Hey guys, I have multiple nested elements within my documents. There's a specific element name that I'm trying to dig in and retrieve data from. This element name exists in several 'sub-documents'. Is it possible to query for this data directly?
[20:57:02] <skram> @kurtis: i believe so, im fairly new to mongo but could you not step through the heirechy with find()?
[20:58:25] <kurtis> skram: Well -- I could use 'dot-notation' and definitely do that. But I'm hoping there's a more generic way to do this. For example "document.*.field-im-looking-for"
[20:58:49] <kurtis> It's sort of hard to explain ... but some Documents have certain fields and others don't
[21:08:14] <crudson> kurtis: no you can't do wildcard matching or xpath-style querying https://jira.mongodb.org/browse/SERVER-736 <- vote for that perhaps
[21:19:06] <belak51> Hi, my IRC client died, so I didn't see if there was an answer, but what's the common way of storing files in GridFS such that the filename and extension can be retrieved? Also what's the common method of storing references to GridFS files in non GridFS records?
[21:38:32] <belak51> ackspony: because if I need to give the file back to the user, they need to know what it actually is
[21:39:11] <belak51> ackspony: so, for referencing the file, would I just store the file's _id in another field of the object?
[21:40:53] <rpcesar> ok, i have a question. I have a working sharded replica set consisting of 6 computers divided over 2 replica sets. I have 2 sharded databases running. What I would like to do is entirely copy one of the databases and produce a secondary database with identical collections and sharding attributes, quite literally clone the database on the same replica set. whats the easiest way to do this?
[22:06:19] <swaagie> is it me or can't you set the user role while doing addUser with the node.js mongodb client?
[22:24:32] <belak51> Is mongodb fault tolerant? I've been seeing a number of posts lately about how it tries, but doesn't really work. Are these true, or just a load of crap?
[22:29:28] <diegows> I have an array of numbers in my docs, and I want to sort the documents by some specific position of the array
[22:29:50] <diegows> the sort works, but Mongo complains that I need a index when I have a big collections
[22:30:04] <bundini> Seeing random crashes and pegged system recently - running 2.2.1. Our logs show thousands of " Thu May 30 00:10:46 [conn777] warning: ClientCursor::yield can't unlock b/c of recursive lock ns: queue.txlog top: { err: "unauthorized" } " messages.
[22:32:06] <bundini> Always with a particular connection as well. In some cases the connections (generally belonging to long running worker processes) eventually succeed (9 hour later in one case): Thu May 30 01:29:43 [conn424115] query queue.schedule query: { id: "QfyHZDozfviVA3HA" } ntoreturn:1 ntoskip:0 nscanned:13850 keyUpdates:0 numYields: 36 locks(micros) r:119108 nreturned:1 reslen:355 115ms
[22:47:15] <bundini> I'm wondering what the err: "unauthorized" means as well. While googling the issue I found similar error messages but that particular attribute seemed unique to our situation.
[23:19:59] <skram> anybody, but i figured it's in bytes. just ran into the documentation page for it