pmxbot IRC Log Viewer

[00:07:57] <gyre007> I see so before I rebuild hekman I need to force reelection...

[00:08:09] <hekman> only on your primary

[00:08:13] <gyre007> I mean before I rebuild primary

[00:08:16] <gyre007> yup

[00:08:18] <gyre007> cool

[00:08:21] <gyre007> cheers hekman

[00:08:37] <hekman> you should automatically elect a new primary if you bring it down

[00:08:43] <hekman> just make sure it happens is what i mean :)

[00:08:47] <gyre007> I see

[00:08:49] <gyre007> ok

[00:20:03] <trevorfrancis> I am looking at building a mongo cluster: We are wanting to use it to store a large volume of simultaneous inserts

[00:20:12] <trevorfrancis> and to perform analytics work on them

[00:20:30] <trevorfrancis> does Mongo have an analytics framework?

[00:32:31] <trevorfrancis> also, how hard is it to add new nodes to a cluster?

[00:52:18] <bartzy> Hey

[00:52:29] <bartzy> I executed find() on the mongo shell

[00:52:32] <bartzy> and got this line

[00:52:38] <bartzy> { "_id" : ObjectId("5137e3fca4c9deee0500000d"), "name" : "Fun", "subcategories" : [ ObjectId("5137e3fca4c9deee05000007"), ObjectId("5137e3fca4c9deee0500000a") ] }

[00:53:02] <bartzy> Well, in the shell the spacing between each ObjectId in "subcategories" is much bigger. Any idea why the spacing is there anyway ?

[05:43:07] <zeroquake> wat server side setup do i need in aws , for storing gps data coming for android phonegap app , using mongodb ??

[06:36:56] <someprimetime_> I'm returning a collection of 10 items that I'd like to loop through and augment an easy date to… Basically something that grabs the created date and converts it to a `X time ago` string… and add it to each item in the collection. What's the best way to handle something like this?

[06:37:17] <someprimetime_> Was thinking a for in?

[07:05:46] <rishabhverma> Hey. I'm trying to install MongoDB on a fresh Ubuntu 12.04 LTS machine. But have been getting this error, "locale::facet::_S_create_c_locale name not valid" repeatedly.

[07:06:00] <rishabhverma> Running locale-gen also doesn't help resolve this.

[07:06:11] <rishabhverma> Any ideas on what should I do in such a case/

[07:06:15] <rishabhverma> *?

[08:30:52] <[AD]Turbo> ciao

[08:56:05] <Katafalkas> Hey, I am wondering which image on amazon ec2 I should use for mongodb. 10gen by default uses standard AMIs. But I do like ubuntu. Also there was some benchmarks and they show that mongodb runs better on ubuntu. Any thoughts on that ?

[09:13:24] <richwol> I've seen a huge company store date/time as a string in the format 2013-03-07 09:10:12, the idea being it's easy to query (for example, to return all records matching a particular day you could do /2013-03-07/. I haven't read much about this technique and the performance of it - does anyone know if it's faster this way than doing a similar query against native date objects? Any advantages/disadvantages would also be appreciated so I can make a well informed d

[09:16:35] <saby> I have a collection containing a field status: {status.algo1: "required", status.algo2: "running", status.algo3: "required"}

[09:16:44] <saby> i.e status has some sub fields,

[09:17:43] <saby> is it possible to send a query such that it sends me any document for which anyone of the status is required?

[09:21:18] <eaSy60> Hello, how could I get the id of the last "inserted" document in mongoshell?

[09:23:21] <Nodex> eaSy60 : you can't unless you set it yourself

[09:23:34] <richwol> EaSy60: http://stackoverflow.com/questions/3611277/how-to-get-last-inserted-id-using-the-js-api

[09:23:49] <richwol> gives you an idea..

[09:24:07] <Nodex> ^^ (setting it yourself)

[09:24:26] <richwol> Yeah, just giving him an example

[09:25:00] <Nodex> with regards to your regex question - the performance is about as bad as it can get

[09:25:33] <richwol> Nodex: Yeah I thought it might be, I can't work out why they've done it this way at all..

[09:25:59] <Nodex> personaly I do this ... timestamp : epoch, ymd : Ymd

[09:26:14] <Nodex> ymd: 20130307

[09:26:41] <Nodex> it's sortable and readable

[09:26:49] <richwol> Oh ok, so you can just match ymd exactly, but you still have the timestamp if needed?

[09:26:55] <Nodex> correct

[09:27:12] <Nodex> if you need it any different then just add the difference you need

[09:27:20] <Nodex> example, Ym (201303)

[09:27:56] <eaSy60> richwol, Nodex: Thanks, I'm just coding a migration script so I don't care how much dirty it is

[09:27:57] <Nodex> some poeople store a date object ... date : {y:2013,m:03,d:07}

[09:28:05] <richwol> sounds good, the only issue for me is that I need to query on different timespans.. for example /2013-03/, /2013-03-06/, /2013-03-06 10/ etc. I guess for that reason I'm best off using the Date object

[09:28:08] <Nodex> it's a little more granular

[09:28:44] <Nodex> I would use range query's for timespans

[09:29:00] <richwol> Ah ok, another interesting approach. My only concern about this is the speed - it's gonna be a big table so whichever approach is fastest is gonna have bonus points for me!

[09:29:07] <richwol> sure

[09:29:30] <Nodex> define big

[09:29:44] <richwol> 20 million records +

[09:30:18] <richwol> The date field was going to make up part of a composite key

[09:31:16] <Nodex> are the records large in size?

[09:31:43] <richwol> Difficult to say, they can vary from about 100 bytes up to about 80k

[09:32:02] <richwol> Odd ones could be a lot more

[09:32:19] <saby> i have a field status containing multiple subfields, like status.algo1: "required" and stuff. Is it possible to get all documents which have any subfield's value as "required" ??

[09:33:00] <Nodex> db.foo.find({status.algo1: "required"});

[09:33:05] <Nodex> db.foo.find({"status.algo1": "required"});

[09:33:06] <Zelest> o/

[09:33:21] <saby> Nodex i have status.algo1, status.algo2, status.algo3

[09:33:23] <Nodex> richwol : if they fit in RAM you should be fine on performance

[09:34:08] <saby> db.foo.find({"status.algo1": "required", "status.algo2":"required"...});

[09:34:14] <richwol> nodex: Awesome. I suppose the other benefit of using the date object is that I can cater for data stored in different timezones.. the other methods are a bit limited for that

[09:34:25] <saby> is there a way to do status* : "required"

[09:34:31] <Nodex> saby : please pastebin a document

[09:34:36] <saby> such that I get all the records for which any status field is required

[09:35:20] <saby> sure Nodex

[09:35:41] <saby> Nodex http://pastebin.com/3iuCqpd3

[09:37:04] <saby> so Nodex I want to keep the find query generic as new subfields could get added in near future so I would like to fetch any doc with a status as required

[09:37:49] <Nodex> you're going to need an $or

[09:38:24] <saby> yes

[09:38:43] <saby> but is there a way of not specifying the individual field names?

[09:39:33] <Nodex> no

[09:42:25] <saby> Nodex any way to get the names of fields available in Mongo?

[09:45:17] <Nodex> you can loop a document I suppose

[09:48:56] <saby> hmmm that's gonna be a problem

[09:48:57] <saby> damn

[09:50:20] <kali> saby: maybe this can help: https://github.com/variety/variety

[09:50:38] <kali> ha. no. sorry

[09:50:40] <kali> forget it

[09:50:59] <saby> ok :)

[11:09:23] <multi_io> when you're storing "1:n relations" using DBRefs, do you commonly store the DBRefs (or simple IDs) on the n side (as a single reference), or on the 1 side (as an array of references)?

[11:09:54] <ron> depends.

[13:19:29] <Zelest> I'm having huge issues with failover and mongo in our replicaset.. :/

[13:19:55] <Zelest> Whenever the master dies, the secondary becomes master, but php-fpm give me "no suitable candidate" even if the master is available and up..

[13:20:03] <Zelest> same happens if the master goes down and comes back up short after

[13:20:09] <Zelest> only way to solve this is to restart php-fpm :/

[14:51:35] <jtomasrl> i have a items document that have nested orders with a created_at field, is it possible to order by that?

[14:51:43] <petto> I guys... I'm trying sort collection result by text field. But it not work with accented character. For exemple (a,b,c,d,v, á)

[14:53:30] <Nodex> jtomasrl : use dot notation

[14:53:56] <petto> hey*

[14:56:21] <petto> db.content.find({_id:/PLACE/,title:{$ne:null}},{title:1,_id:0}).sort({title:1}); => {"title":"a"}, {"title":"b"}, {"title":"c"}, {"title":"á"}

[15:02:05] <kali> petto: yes, sort is strcmp based. nothing fancy happening

[15:04:29] <petto> kali: =/ How can I solve this? Or how can I improve this sort?

[15:06:05] <kali> petto: the only thing you can do is normalize the "title" by discarding the accents (and maybe case too). that will improve the situation for some languages (not all of them)

[15:09:42] <petto> oh my god!

[15:10:09] <strigga> mmmh?

[15:23:53] <kali> petto: you may want to watch/vote https://jira.mongodb.org/browse/SERVER-1920

[15:31:03] <Nodex> or just use english hahahahah

[16:05:12] <qhartman> I see that the method to increase the size of a capped collection in 1.x is to create a new larger collection and then copy the contents of the current one over to it.

[16:05:39] <qhartman> Is that the same process in 2.x, or has the feature I see talked about online about a more direct way to do that been implemented?

[16:16:00] <coogle> Does anyone know what this error means?

[16:16:00] <coogle> Couldn't pull any geometry out of $within query:

[16:16:06] <coogle> Couldn't pull any geometry out of $within query: { 0: [ -73.98085499999999, 40.75547299999999 ], 1: 0.0252321865809176 }

[16:17:23] <coogle> never mind... it's because i didn't have a 2d index

[16:25:59] <bartzy> Hey

[16:26:19] <bartzy> when I encode a MongoId (PHP driver) to JSON via json_encode, the result is :

[16:26:46] <bartzy> {"$id":"5137e3fca4c9deee05000001"}

[16:26:58] <bartzy> This is not expected - it should be just the string, not an object like that.

[16:27:30] <bartzy> What about MongoId implementing JsonSerializable ?

[16:32:24] <Derick> bartzy: file a feature request please - however, I don't quite agree it should just be a string, as it's a bit more than that

[16:32:29] <Derick> why are you turning it into JSON?

[16:32:40] <bartzy> Derick: To forward it to our JS app.

[16:32:52] <bartzy> our JS app need to know the id of that item/object/whatever...

[16:33:03] <bartzy> and that's our representation for the id (the ObjectId of the document)

[16:33:29] <bartzy> and now instead of going to photo.id (in JS for example), it needs to go to photo.id.$id , which is weird...

[16:33:35] <Derick> we do still have to support php 5.2 though.. and jsonserializable is 5.3 (or 5.4 even)

[16:34:07] <bartzy> Derick: I don't understand the reasoning behind having a json like that: {"$id":"5137e3fca4c9deee05000001"}

[16:34:23] <bartzy> It's not that json_decode would bring back MongoId to "life", it would just be stdClass

[16:34:28] <Derick> it's the same as vardump... so it makes sense if you look it like that

[16:34:44] <Derick> I do indeed not like the "$id" name though, as it's confusing

[16:34:51] <Derick> (and difficult to use in JSON)

[16:35:10] <bartzy> so most users that are just doing json_encode($results) from a mongo query to their JS app - need to go over each document and cast the MongoId to string ?

[16:35:30] <Derick> well, you just said you can do photo.id.$id yourself...

[16:35:54] <Derick> we can't really change behaviour either :-/

[16:36:16] <Derick> I think in the end, it would be best if you can specify your own classes instead of MongoID/MongoCursor etc, so that you can modify behviour yourself

[16:37:50] <bartzy> Derick: Wouldn't that be pretty slow ?

[16:38:11] <Derick> maybe, but atleast we don't have to make a decision

[16:38:34] <Derick> implementing jsonserializable now, and changing it to a string for that will break BC, which we can't do before 2.0 (and that we'd like to avoid)

[16:48:33] <fredix> hi

[16:48:49] <fredix> it seems that mongo::ScopedDbConnection can connect to a single mongodb instance, right ?

[17:09:36] <bartzy> Derick: when 2.0 will arrive ? Why you can't do that before 2.0 ?

[17:09:49] <Derick> we can not break people's code

[17:09:58] <Derick> changing behaviour means breaking people's code

[17:10:02] <Derick> 2.0 is not planned

[17:19:16] <bean__> Backwards compat matters

[17:51:48] <jiffe98> why would show dbs not show all the available dbs?

[17:51:56] <jiffe98> I'm doing this through the router

[17:52:16] <jiffe98> it isn't listing my mail db which exists and when I use it, it shows me the collections in it fine

[18:27:38] <dmilith> hi

[18:27:43] <dmilith> I want to build mongo from source

[18:28:01] <dmilith> on normal systems like darwin/ freebsd with clang I'm adding --clang to scons

[18:28:06] <dmilith> and it works fine

[18:28:21] <dmilith> but documentation lacks on information of how to give explicit path to C++ compiler

[18:28:33] <dmilith> cause it's ridiculous to build mongodb with g++

[18:28:37] <dmilith> any hints?

[18:30:18] <dmilith> note: on linux I have clang in custom path

[18:30:31] <dmilith> (it's in PATH, but it's scons..)

[18:31:34] <wereHamster> scons --cxx=/path/to/...

[18:40:00] <dmilith> ah

[18:40:03] <dmilith> --cxx

[18:40:24] <dmilith> It's really fucking awesome to know. Cause no documentation ever mentioned such option.

[18:42:10] <wereHamster> not even scons --help?

[19:06:15] <dmilith> wereHamster:

[19:06:16] <dmilith> src/mongo/db/namespace_details.h:112:22: error: private field 'reserved2' is not used [-Werror,-Wunused-private-field]

[19:06:16] <dmilith> unsigned reserved2;

[19:06:30] <dmilith> and how to deal with such ridiculous "errors"?

[19:07:26] <dmilith> allright. I see

[19:09:03] <dmilith> no, I see only --ignore-errors

[19:09:07] <dmilith> and this is not same

[19:13:27] <dmilith> --warn=WARNING-SPEC

[19:13:31] <dmilith> what does it means?

[19:19:28] <dmilith> no documentation at all..

[19:19:36] <dmilith> it's real open source project.

[19:19:47] <dmilith> I tried even read SConscript

[19:19:53] <wereHamster> you can get your full refund back if you don't like it.

[19:19:55] <dmilith> no fun at all

[19:19:59] <wereHamster> even with interest!

[19:20:02] <dmilith> I don't use it

[19:20:32] <dmilith> and never wont, but I need to build it for some other stupid people who will be

[19:21:16] <wereHamster> cna'ot' you use a binary package?

[19:22:43] <dmilith> nope

[19:22:52] <dmilith> I'm using very ancient distro

[19:23:06] <dmilith> I really want only to get rid of those warnings as errors

[19:23:16] <dmilith> and I got lost and never back, I promise

[19:23:43] <dmilith> --warn=WARNING-SPEC, --warning=WARNING-SPEC

[19:23:43] <dmilith> Enable or disable warnings.

[19:23:48] <dmilith> what does it means?

[19:23:58] <dmilith> where can I find 1 word about it

[19:24:13] <wereHamster> use the source, luke

[19:27:09] <dmilith> yea, sure

[19:27:18] <dmilith> --clang --cxx='${SOFTWARE_DIR}Clang/exports/clang++ -w'

[19:27:35] <dmilith> is the solution, but i bet you wont even mention about it in your great fucked up site

[19:27:49] <dmilith> this project is so lame

[19:42:38] <SteveO7> Does anyone know if mongoimport can be used to update fields in an embedded document?

[19:56:21] <ehershey> what distro are you using?

[19:56:38] <ehershey> (dmilith)

[19:57:03] <dmilith> ehershey: debian stable

[19:57:18] <ehershey> debian stable is an ancient distro?

[19:59:32] <dmilith> it is

[20:00:05] <ehershey> I can't keep up

[22:15:10] <redsand> is it possible to add a new shard to an existing set if the set already has authentication enabled?

[22:38:13] <toxster> hi, i am setting a sharded cluster, do i need to have mongod running on each server? now mongos complains that it's using the same port, or does mongos tae over what mongod does?

[22:39:15] <toxster> should i just kill all mongod processes?

[22:42:29] <julian-delphiki> toxster: you'll want to change what port your mongod is listening on.

[22:43:01] <toxster> julian: ok do i need the original mongod daemons

[22:43:18] <julian-delphiki> yep

[22:43:23] <toxster> ok

[22:43:51] <toxster> what host do i direct my client at, and to which process? mongos, mongod configsvr, or original mongod

[22:44:17] <julian-delphiki> toxster: http://docs.mongodb.org/manual/core/sharded-clusters/#infrastructure-requirements-for-sharded-clusters

[22:45:11] <julian-delphiki> it'd be the mongos

[22:45:14] <fredix> hi

[22:45:25] <julian-delphiki> since, you know, the mongos is the server that controls the shards, etc

[22:45:33] <toxster> so m understanding is that i should direct my client towards a mongos host and port

[22:45:41] <fredix> what should the connection string to connect to a replica set with ScopedDbConnection ?

[22:47:09] <fredix> ScopedDbConnection *replicaset = ScopedDbConnection::getScopedDbConnection("yourreplicasetname/ip1,ip2,ip3" ); is it ok ?

[22:47:43] <julian-delphiki> toxster: yes, and the mongos should handle the rest transparently

[22:47:45] <julian-delphiki> i believe

[22:48:25] <toxster> ok, so i'll just change mongos's port and update my client code to connect to that port

[22:48:36] <toxster> all should be well?

[22:49:19] <toxster> only 1 mongos should be needed when i test right, the other is just for failover

[23:28:11] <toxster> weird, when i run mongos on port 27020, mongod --configsvr on 27018, and 10gen mongo 2.2.3 on 27017, when i start mongod on 27017 i get Fri Mar 8 00:20:15 got signal 2 (Interrupt), will terminate after current cmd ends

[23:33:12] <toxster> seemed to be that the configsvr has t be started after mongod

Log file Viewer

Help | Karma | Search:

#mongodb logs for Thursday the 7th of March, 2013