pmxbot IRC Log Viewer

[00:37:16] <timeturner> anyone ever try making a regex full text search system with mongo

[00:37:53] <_m> Negative. Have simply leveraged ElasticSearch.

[00:38:18] <timeturner> it just feels like a waste though

[00:38:30] <_m> Depends on what your use-case is

[00:38:34] <timeturner> copying all the indexes into a search engine that looks exactly like mongo

[00:38:41] <timeturner> I mean is it worth it?

[00:38:50] <_m> Obviously.

[00:39:39] <timeturner> I'm just stumped as to why 10gen doesn't optimize mongo for full text search or at least include a separate tool for that

[00:39:41] <_m> One could spend hours/days/months implementing a fuzzy-match search with weighting.

[00:39:56] <_m> Because a database is not a search engine.

[00:39:56] <timeturner> yeah

[00:40:08] <_m> KISS methodology

[00:40:09] <timeturner> I figure I'll have to go with elasticsearch as well

[00:40:15] <timeturner> for lack of better alternatives

[00:40:20] <timeturner> yup

[00:40:33] <timeturner> how did you get started with elasticsearch anyways?

[00:40:46] <_m> I hated working with Sphinx/Solr

[00:40:59] <timeturner> ah, yeah that'll do it haha

[00:41:09] <timeturner> in your experience does elasticsearch take a lot of resources while it is running?

[00:41:10] <_Tristan> sphinx is much less of a pain now that it has sphinxql

[00:41:44] <_Tristan> before that, it was pure agony

[00:43:17] <_m> timeturner: Our production box is using about 2gigs of RAM

[00:43:23] <_m> And has a load like htis: oad average: 0.89 0.80 0.67

[00:43:32] <_m> So, no… not really.

[00:44:26] <_m> I should mention that's a quad-core box with 4gigs of ram. Pretty lean, IMO

[00:45:15] <_m> _Tristan: Agreed. But that's the same reason I don't use puppet. wtf should I learn a not-quite-ruby DSL? (granted this is no longer the case for pupper)

[00:45:18] <_m> *puppet

[00:46:04] <_Tristan> I know, it's like, I finally got away from SQL for good, and now my search engine makes me use it

[00:47:52] <_m> Seriously.

[00:48:03] <_m> ElasticSearch is all JSON all the time. That makes me pretty happy.

[00:48:30] <_Tristan> orly

[00:48:32] <_m> Also the JSON-based configuration. And not needing to restart the instance for config/index/other changes to be picked up

[00:49:22] <_m> timeturner: If you're using Ruby, we released an OSS driver.

[00:57:34] <_m> https://github.com/edmundsalvacion/elasticsearch-ruby

[04:13:41] <niriven> hi, i have been struggling for some time on how to design a database. I come from standard relational databases so i have a tendency to do want to do things that may be wrong.

[04:16:17] <niriven> I have two types of documnts, users, events. user and event has a shard identified, user_id. users is small (10000 or so), and events is large (150 million) for a test set. the event may or may not have an id is a valid id for a user (it can be null). The majority (95% of the in event) is a null user_id, and most of the queries i will be running require one for me to be interested in the result.

[04:17:46] <niriven> so do i go about 1) having a db for users, and a db for events, and "join" in my code, or create one big event docoment which replicates user data per event, or have one big user document which for each user has its associated events (but given that the majority of data does not have a user means i would have to assign it to a fake user, making a massive document (gigs in size)). Anyone have any insight?

[04:19:21] <crudson> do you ever care about the events with no user?

[04:20:12] <crudson> (I think you did answer that, re-reading)

[04:20:27] <niriven> nope

[04:20:42] <niriven> eventually events might be updated with a a valid user, but the queries only care if it has a user

[04:20:53] <crudson> is it worth even capturing those then?

[04:21:14] <niriven> good question, i'm told it is :)

[04:22:11] <crudson> and I expect you're not going to always want every event for a user.

[04:22:16] <crudson> :)

[04:23:08] <niriven> correct :)

[04:23:24] <niriven> under other filter critera, of course

[04:24:03] <niriven> eg. user must match crit 1, 2, 3, and relate to a event that that has critera a, b, c

[04:25:19] <crudson> I think 1) is the only reasonable solution. Replicating user data across events seems like too much work and redundancy. Having gigs in size document rules that out from the very start, both technically for mongodb constraints and logically.

[04:25:25] <niriven> ideally one document with users / events if all these events had users :(

[04:26:40] <niriven> crudson: great, i came to that conclusion as well, though not an ideal situation, it should work

[04:26:45] <crudson> making 1 extra user query for a v large number of event queries is nothing

[04:26:57] <crudson> the only issue I would see is a matter of concerrency

[04:27:05] <niriven> crudson: much appreciated!

[04:28:01] <niriven> lastly, how much should i start to nitpick about redundancy in small documents, lets say i have 10 fields in a document with 1 level, basic object. 4 of those 10 fields are repeated often, so i start to break apart? im assuming not

[04:29:01] <niriven> if i break it apart it creates problems queries, if i dont i may have increased size on disk

[04:31:25] <crudson> it's a matter of reads vs writes. The logical integrity of your data is the important thing. If one gets updated a lot, which would result in having to update a ton of other documents, I would consider that a warning.

[04:31:57] <crudson> e.g. having a zip code, and always adding geo coords or state, that is not a risk, as that (almost never) changes

[04:32:15] <crudson> but the benefit of having lng/lat in that instance is a real benefit

[04:32:23] <niriven> most likely those fields are not updating often, its just insert and forget, and query later

[04:32:24] <crudson> even if you have that in your zipcode collection

[04:32:57] <niriven> gotcha

[04:33:42] <niriven> this document can have any set of values for its 10 attributes, i'm just worried there might be replication across those attributes, nothing obvious that can be represented as its own document though

[04:34:07] <crudson> if you do have v large documents though, judiciously use 'fields' or 'slice' to restrict what portions of the documents you are returning across the wire.

[04:34:37] <niriven> gotcha :)

[04:34:39] <niriven> thanks crudson

[04:35:49] <crudson> it's a pleasure.

[04:56:54] <niriven> i'm also debating wheter or not to create a compound index on a document. i read on mongodb that if you index a,b,c you it is only efficent to query on a, a,b or a,b,c, not a,c, or c,b?

[05:00:35] <niriven> given i could a query that filters on [a,b,c], [a,b], [a], [a,c], [b,c], is it best to create a compound index or individual indexes?

[05:02:40] <niriven> nevermind, googled helped find a good example :)

[05:54:06] <VooDooNOFX> Need some advice on a fresh mongo installation (mongo-10gen-2.2.0-mongodb_1). Inserted 10,897,335 entries. Looking up the process in top shows the following: 29641 mongod 16 0 24.6G 4046M 4016M R 43.0 51.5 0:55.67 /usr/bin/mongod -f /etc/mongod.conf. That's 24.6G SWAP usage?

[06:22:31] <telmich> good morning

[06:23:19] <telmich> I would like to run an arbiter for a replica set externally of our 2 datacenters - now when I add it to the replica set, it tries to connect to the internal host names, which are not reachable from external

[06:23:57] <telmich> is there any way the arbiter gets contacted _from_ the other nodes instead of connecting to them?

[07:26:57] <[AD]Turbo> hola

[07:28:22] <wereHamster> VooDooNOFX: isn't it 24G virt?

[07:36:54] <Null_Route> Hi Guys! in http://docs.mongodb.org/manual/release-notes/2.0/ , "Upgrade all mongos instances first, in any order. Since config servers use two-phase commit, shard configuration metadata updates will halt until all are up and running."

[07:37:03] <Null_Route> do they mean "config" instances ?

[07:37:58] <Null_Route> The next line in the config says "Upgrade mongos routers in any order."

[07:55:33] <kkszysiu_work_> Hello :)

[07:57:01] <kkszysiu_work_> And have indexes already

[07:58:10] <kkszysiu_work_> But quering it using db.data.find() took loong time, especially when we're making lot of requests

[07:58:47] <kkszysiu_work_> So if we run 4 queries in parallel, mongo seems to queue them

[07:58:55] <kkszysiu_work_> And calling one-by-one

[07:59:50] <kkszysiu_work_> Is possible to configure mongo to call for example 2 requests in parallel?

[08:00:15] <kkszysiu_work_> Because its not great if one request blocks another...

[08:02:48] <NodeX> how long are your queries taking?

[08:06:00] <kkszysiu_work_> NodeX: 3 seconds

[08:06:03] <kkszysiu_work_> 3-4

[08:07:12] <NodeX> can you do an explain() on a query and pastebin it

[08:16:52] <NodeX> I guess not

[08:18:11] <Null_Route> hmmm

[08:24:04] <kkszysiu_work_> NodeX:

[08:24:05] <kkszysiu_work_> http://pastebin.com/FxQWtFA2

[08:24:29] <kkszysiu_work_> Its fast

[08:24:37] <kkszysiu_work_> Maybe sending data here is a bottleneck

[08:32:55] <NodeX> ;)

[08:34:01] <NodeX> range queries also have to be the last part of an index iirc

[08:40:06] <Null_Route> Anyway - Upgrading a sharded Replica Set - Mongos first, or config servers?

[08:40:25] <Null_Route> http://docs.mongodb.org/manual/release-notes/2.0/ is unclear

[08:47:19] <Lujeni> Null_Route, your driver at first

[08:47:24] <Lujeni> then Upgrade all mongos instances first, in any order.

[09:23:53] <Lujeni> Hello - I meet this error ( assertion: 13111 field not found, expected type 2 ) when i try to restore. However, the source and destination are the same version.

[09:27:38] <arussel> when modifying a document with .update({}, {$set: {foo: "myvalue"}}, how can I set using a field from the document ?

[09:27:56] <NodeX> not possible

[09:28:42] <IAD> through "command" may be

[09:28:50] <NodeX> it's NOT p;ossible

[09:29:14] <arussel> so I have to read the doc, modify the doc and save it ,

[09:29:18] <arussel> s/,/?/

[09:29:24] <NodeX> (not in one query) - you have to do it in 2 queries

[09:29:46] <NodeX> ^^ correct

[09:32:17] <arussel> NodeX: something like this: http://pastie.org/4752559 ?

[09:34:18] <NodeX> something along those lines yup

[09:35:46] <arussel> mongo is not happy with the last line: ("assert failed : need an object")@shell/utils.js:46

[09:36:20] <NodeX> recipie needs to be an object ...

[09:36:39] <NodeX> db.recipe.update(updateCode, newData);

[09:37:25] <NodeX> var updateCode={foo:'bar'}; var newData={$set:{bar:'baz'}};

[09:38:14] <arussel> here you set bar to baz when foo is bar

[09:38:50] <arussel> I'm trying to update each doc to a value somewhere else in the doc

[09:39:51] <NodeX> it's an example

[09:40:01] <NodeX> db.recipe.update(updateCode, newData); <------------- that's the point of it

[09:40:38] <arussel> how do I define newData in term of the document being modify ?

[09:41:40] <NodeX> {$set:{bar:'baz'}

[09:41:51] <NodeX> {$set:{bar:cookie.foo}

[09:42:30] <NodeX> that will set bar to the value of cookie.foo ... {cookie : { foo : 123 }} ... 123 in this case

[09:43:01] <unknet> can i ask here about mongoid?

[09:44:15] <NodeX> you can ask unknet

[09:45:19] <unknet> NodeX, I have a model with an embedded_in relation and a belongs_to relation

[09:45:46] <unknet> but it seems it is not alowwed and i dont understand why

[09:50:59] <arussel> NodeX: second try (still unsuccessfull) http://pastie.org/4752559

[09:53:03] <NodeX> with what error

[09:54:53] <boll> Is there any word on whether 2.2.1 is on track to be released according to the roadmap?

[09:56:10] <arussel> NodeX: no error, but no 'secondTitle' field either

[10:00:13] <arussel> NodeX: forget about it, I made a typo

[10:00:18] <arussel> thanks for your help

[11:46:30] <tonny> Hey, i have a little issue, don't know if it is related to gridfs or nginx-gridfs module, i have setup a gridfs with more than 1000000 image files, and serve the files with nginx with gridfs module. every webserver i have use an mongos application server. everything is fine but once in a while a request on 1 webserver hangs and timesout. and won't serve till i restart the mongos on that webserver. But the same file works on a different webs

[11:46:59] <tonny> Maybe that one of you has or knows this problem and how to fix it ?

[12:20:05] <gyre007> why am I getting following exception ? replSet exception loading our local replset configuration object : 13132 nonmatching repl set name in _id field

[12:20:07] <gyre007> uhh

[12:37:27] <unknet> Hi

[12:50:44] <Lujeni> gyre007, can you paste your config ?

[13:06:49] <Vinx> hello all

[13:07:12] <Vinx> i have a very difficult question about mongodb....

[13:07:17] <Vinx> is there soem one?

[13:08:15] <aboudreault> just ask, and be patient

[13:09:33] <Vinx> I like to embedd the MongoDB server into my own C++ application.....

[13:10:50] <Vinx> there is any "built in" solution.. i would like to avoid to get the source doce of mongdb.. and "manually" wrap the core..

[13:11:24] <vegivamp> helo

[13:11:38] <aboudreault> built in solution for what?

[13:11:46] <NodeX> Vinx: it's not possible

[13:12:32] <vegivamp> I seem to have a bit of a fight with mongobackup (2.2) - I can't seem to perform a backup of anything but the admin DB using a read-only admin user. Is this expected behaviour? Write privileges for a read-only operation seem overbearing :-)

[13:12:54] <vegivamp> s/mongobackup/mongodump/

[13:14:59] <Vinx> #NodeX i know it is a very complex question, because of the distrubuited nature of mongodb I assume..

[13:15:35] <Vinx> aboudreault: to embedd mongodb server in a c++ application

[13:16:03] <aboudreault> you can't

[13:16:15] <aboudreault> I don't see the use either

[13:16:22] <aboudreault> it just add extra maintenance to your app.

[13:18:04] <Vinx> I have some production limitation.. that does not allow me to install no server on the production machine.. so i am trying to use some tricks.. for instance what do you all think about: http://stackoverflow.com/questions/6437226/embedded-mongodb-when-running-integration-tests

[13:18:18] <algernon> not possible.

[13:18:34] <Vinx> ok

[13:18:36] <Vinx> :)

[13:19:24] <algernon> set up a mongodb instance for testing, use that from your app when running CI. problem solved.

[13:19:52] <aboudreault> Vinx, maybe you should just use look another DB, maybe a file-based DB ?

[13:20:01] <tonny> hi all, i run into some issues with gridfs, you seen my post while back ?

[13:21:38] <tonny> Hey, i have a little issue, don't know if it is related to gridfs or nginx-gridfs module, i have setup a gridfs with more than 1000000 image files, and serve the files with nginx with gridfs module. every webserver i have use an mongos application server. everything is fine but once in a while a request on 1 webserver hangs and timesout. and won't serve till i restart the mongos on that webserver. But the same file works on a different webs

[13:21:44] <Vinx> aboudreault i see.. d

[13:21:54] <Vinx> I like mongo due to json docum,ents...

[13:22:11] <Vinx> do you have any suggestion as alternative to mongodb?

[13:22:22] <NodeX> couchdb uses json style

[13:22:23] <aboudreault> yeah. then ask your damn sysadmin to install that great soft :)

[13:22:24] <Vinx> ( that could be embedded into a c++ app? )

[13:22:32] <NodeX> perhaps you can embedd that?

[13:22:39] <cedrichurst> couch is erlang-based

[13:22:50] <aboudreault> will be tough yeah

[13:23:00] <Vinx> couchdb.... mmm any comparison with mongodb?

[13:23:05] <cedrichurst> plenty

[13:23:18] <cedrichurst> http://www.mongodb.org/display/DOCS/Comparing+Mongo+DB+and+Couch+DB

[13:23:31] <NodeX> http://couchdb.apache.org/ <--- also

[13:23:35] <cedrichurst> though it's mostly on the 10gen side, and written when mongodb was the scrappy upstart

[13:24:31] <Vinx> thanks guys...

[13:24:47] <Vinx> it is a very interesting topic anyway...

[13:32:55] <vegivamp> So, anyone else ever tries mongodump with a read-only user?

[13:34:17] <Rhaven> hi all, i got this in config server log

[13:34:18] <Rhaven> [DataFileSync] flushing mmaps took 60105ms for 3 files

[13:34:46] <Rhaven> and after this my cluster died

[13:36:26] <seltar> hey.. how would i go about finding all elements in a collection that are unique, without knowing what they are? so skip all the duplicates, basically

[13:36:50] <NodeX> distinct?

[13:37:09] <seltar> probably, i'll look into that, thanks

[13:37:36] <NodeX> it will bring you back an array of the distinct not the documents

[13:38:15] <seltar> well, not exactly what i wanted

[13:38:53] <NodeX> you can use the aggregation framework to do it

[13:39:41] <seltar> ok, thanks for hte helt

[13:39:44] <seltar> the help*

[13:47:15] <vegivamp> Really? Nobody ever makes backups here? :-)

[13:49:09] <wereHamster> vegivamp: why would you think that?

[13:50:44] <NodeX> I backup every hour lol

[13:52:08] <vegivamp> using mongodump ?

[13:52:32] <vegivamp> wereHamster: because of the deafening silence re: my question towards experiences

[13:53:29] <vegivamp> As said before, I'm encountering problems backing up with a read-only admin user

[13:53:51] <vegivamp> yet none of the docs I can find mention anything at all re. what permissions are required for backup

[13:54:21] <vegivamp> and plain old logic suggests that backup is a read-only operation

[13:56:17] <NodeX> but the problem is easily fixed

[13:57:18] <vegivamp> You may understand my hestiation at putting all-powerful credentials in a script.

[13:59:45] <NodeX> what does it matter of you control who executes the script

[14:00:37] <vegivamp> It is an extra layer of security.

[14:02:00] <NodeX> I imagine that mongodump has to be able to write to lock things when backing up so perhaps add a read/write user for backups?

[14:02:18] <vegivamp> Why can it perfectly back up the admin db, then?

[14:03:10] <NodeX> pass

[14:03:38] <NodeX> that's the "imagine" part

[14:05:26] <vegivamp> Hmm, no, I seem to remember that incorrectly; it can't backup the admin db either

[14:05:31] <vegivamp> might well be a locking issue, then.

[14:05:36] <vegivamp> Seems rather silly, though.

[14:30:05] <vegivamp> According to O'Reilly's definitive guide, "mongodump uses the regular query mechanism" and "is no guarantee of consistent backups" - so it seems particularly silly that a read-only user would not be able to perform queries.

[14:30:32] <NodeX> I would think the consistent bit would be due to safe writes

[14:30:43] <NodeX> or writes still in memory waiting to be flushed

[14:30:48] <vegivamp> indeed

[14:31:17] <vegivamp> but if it's using the regular query mechanism, there's no obvious reason why a read-only user would not be able to backup

[14:31:30] <vegivamp> unless a read-only user is quite useless in it's entirety :-)

[14:32:26] <NodeX> are you dumping everything?

[14:33:45] <vegivamp> that's what I'm trying to do, yes

[14:34:00] <vegivamp> it looks more and more as if I'll just have to roll my own script with mongoexport

[14:34:37] <vegivamp> although apparently "Neither JSON nor TSV/CSV can represent all data types. Please be careful not to lose or change data (types) when using this. For full data fidelity, or backups, please use mongodump."...

[14:36:36] <NodeX> You can just take a snapshot if you use journals / lvm

[14:36:51] <vegivamp> with fsync/lock, yes

[14:37:00] <vegivamp> I might be relegated to that

[14:37:36] <vegivamp> heeh. Damn admins didn't set it up using lvm.

[14:38:01] <NodeX> journaling should be enabled by default post 2.0

[14:38:20] <NodeX> I dont like full backups as they are Huge so I increment mine

[14:38:44] <vegivamp> is there documentation of that? You just copy the journals out, then?

[14:39:07] <NodeX> I saw some yetserday when I was researching backing up gridfs, one sec let me dig it out

[14:40:42] <NodeX> http://docs.mongodb.org/manual/administration/backups/#block-level-backup

[14:47:17] <vegivamp> Hmm. That merely copies the snapshot

[14:47:34] <vegivamp> doesn't that simply expand into a full disk image anyway? How is that differential?

[14:48:35] <NodeX> http://benjamin-schweizer.de/dsnapshot.html

[14:48:47] <NodeX> maybe that can help - you can snapshot your data dir

[14:49:06] <coopsh> I don't get the semantics of 'primaryPreferred' read preference in 2.2

[14:49:09] <NodeX> sorry you aint got lvm

[14:49:41] <vegivamp> Oh, I can get LVM if necessary, that's not the problem. I just fail to see how snapshot backups yield off-host differential backups

[14:50:05] <coopsh> tried it out, but mongos returns an exception: 'no master found on ...'. primaryPreferred is used and secondaries are available. Works as designed?

[14:53:26] <IceGuest_77_> hi, i have an application where i can use a text as kind of contract rules... actually a contract :)... happens that in the contract we have some calculations like d=c+e*q and q has a previous set .. i have some examples of the contract, i need some advices

[14:59:00] <NodeX> advice on what?

[15:00:56] <IceGuest_77_> i would like you to see my text and see if it fits ok to mongo

[15:01:05] <IceGuest_77_> i can paste it,

[15:01:10] <NodeX> pastebin

[15:01:27] <IceGuest_77_> yes, sure.. brb

[15:06:18] <IceGuest_77_> http://pastebin.com/xs2UA5Be

[15:07:31] <NodeX> that's just a page of text, I am not sure what you're asking

[15:07:41] <IceGuest_77_> i can make changes to the "contract" actually i have designed this way because i think i can teach emploees to use this syntax to write contracts

[15:07:46] <IceGuest_77_> let me explain:

[15:08:37] <IceGuest_77_> fechaVigencia 01/01/2000 10.00.00

[15:08:37] <IceGuest_77_> prestador PMC

[15:08:37] <IceGuest_77_> obraSocial OSDE

[15:09:05] <NodeX> and you want the rendered version to sit in mongo?

[15:09:06] <IceGuest_77_> this is a contract between prestador PMC and obrasocial OSDE

[15:09:14] <IceGuest_77_> yes i do

[15:09:50] <IceGuest_77_> but in the calculation area ok.. you must do the calculations using this litle scripts

[15:10:09] <NodeX> then in your scripting language get the size of the final string

[15:10:37] <IceGuest_77_> maybe

[15:11:33] <NodeX> there is no maybe, that's the only way to do it

[15:29:46] <coopsh> is anybody actually using primaryPreferred read preference?

[15:30:15] <coopsh> I've the feeling that I'm the first one testing that feature ;)

[15:31:05] <doxavore> Are there any tools to see why MongoDB is only using a fraction of available physical memory and instead running about 50-80% disk IO?

[15:32:12] <doxavore> (Ideally something that works with the 2.0.x branch)

[15:33:55] <coopsh> doxavore: how do you know that only a fraction of physical memory is used?

[15:35:41] <doxavore> coopsh: It's hovering at ~40% system memory (nothing is in swap). Is there something else I should be looking at?

[15:36:31] <coopsh> system memory? which OS? what's the cache size?

[15:36:46] <coopsh> free -m is your friend on linux

[15:37:18] <doxavore> ahh - it's there, in the cache, not actively used by mongodb. gotcha.

[15:37:50] <coopsh> right. see http://www.mongodb.org/display/DOCS/Caching

[17:53:51] <jrdn> what's the right way to check if an aggregation is using the index properly?

[17:53:55] <jrdn> just by doing a find?

[18:03:51] <ron> I've officially lost trust in morphia.

[18:07:12] <jrdn> also, i've got a lot of data which has a created_at mongo date field…. how can I query something like, "hours 10-15 of the last 7 days"

[18:08:10] <jrdn> i know I can rework my data to have "date, year, month, day, hour, minute" fields… but just wondering if i can do it with one indexed field

[18:42:28] <eherot> Is there a way to specify which server in a replica set acts as the data source when a new member is added to the set (assuming slaveOk = true)?

[18:52:40] <underh2o> Need help setting up mongodb with a Acquia Drupal Dev Stack configuration on a macbook. Can anyone help?

[18:53:22] <Derick> underh2o: you need to ask questions before help can be given

[18:53:36] <Forbeck> hi, can I set table scan per collection?

[18:56:10] <underh2o> Derick: I guess my question is Does anyone have experience with how you set up mongodb to work with Acquia Dev Stack? Would you mind sharing the process with me?

[18:57:26] <Derick> I've done it once :-)

[18:57:31] <Derick> I doubt that counts as "Experience" though

[18:59:27] <underh2o> I have mongo working on a macbook. I have Acquia on the same macbook. I have not been able to get the mongo-php driver to install properly to work with the Acquia Dev Stack. I think they use there own PHP with Acquia. How did you get the driver/db to work with the stack?

[18:59:52] <Derick> I didn't use the acquia php stack

[19:08:45] <underh2o> Derick: What did you use?

[19:12:18] <doxavore> When an RS secondary isn't being queried from directly, the only thing causing it to have high page faults are inserts/updates, right?

[19:27:57] <jY> doxavore: yes

[19:28:26] <jY> probably just updates

[19:33:28] <Derick> underh2o: I compile my own PHP

[19:33:34] <Derick> (but I'm not on a mac)

[19:34:13] <Aartsie> Hi Derick how is your trip ?

[19:36:43] <Derick> Aartsie: so far so good

[19:36:48] <Derick> I'm at UG #2 - orlando

[19:37:19] <Aartsie> Derick: Nice!

[20:28:09] <joe_p> after adding new member to replica set getting assertions. [slaveTracking] User Assertion: 11000:E11000 duplicate key error index: local.slaves.$_id

[20:29:02] <joe_p> hoping not to have to stop primary as many production applications run on that machine

[20:36:20] <lukebergen> quick question. I feel like this should be easy but I'm not quite sure how do do this

[20:36:49] <lukebergen> if I have an array name = ["alice", "joe", "bob"]

[20:37:12] <lukebergen> and a collection of documents that all have a "names" attribute (I guess in this weird example people can have multiple names)

[20:37:46] <lukebergen> how would I query mongo to get me all the people whose name array contains at least one of the elements from my name array variable?

[20:38:23] <lukebergen> Is that possible in a single query or would I just have to do something like People.find({names: name[0]}) + People.find({names: name[1]}) + ...

[20:39:38] <crudson> lukebergen: use http://www.mongodb.org/display/DOCS/Advanced+Queries#AdvancedQueries-%24in

[20:40:16] <lukebergen> I saw that but wasn't sure if it'd work if the field in the document is also an array

[20:41:39] <crudson> did you try it?

[20:42:35] <crudson> trying it in the mongo shell would have taken less time than typing your long question here ;)

[20:43:37] <lukebergen> heh, probably

[20:43:53] <lukebergen> yeah, that seems to do what I need it to

[20:44:06] <crudson> cool

[21:10:52] <rustyrazorblade1> newrelic is reporting a lot (hundreds) of calls to find_one, but we don't have that anywhere in our codebase. does pymongo use find_one internally as a next() operation on the resultset?

[21:25:30] <TTimo> interesting. didn't realize a replica set would need > 5GB of disk out of the gate, even if there is no data in the db yet

[21:57:52] <xaka> i'm trying to update document using $set and it does nothing: db.collection.update({_id:"505a396f8cdaf26525000061"}, {$set:{field:1}}). "field" is always 0

[22:04:45] <cmendes0101> xaka: have to ask. Can you pull up the record with db.collection.find({_id:"505a396f8cdaf26525000061"}) ?

[22:08:45] <xaka> cmendes0101: no, but what i've found is that it works when i wrap _id into ObjectId

[22:09:10] <cmendes0101> yah in CLI you need that

[22:09:33] <cmendes0101> did that also correct your update?

[22:10:32] <xaka> yeap :)

[22:13:17] <VooDooNOFX_> I'm seeing something like 332 ops/sec 1 MB/sec results when I run mongoperf according to the wiki page instructions. Any particular reason its much lower than the example output? (core i3, 1tb sata hdd, 8gb ram).

[22:14:33] <VooDooNOFX_> while on my i7, 16gb ram, osx box i'm seeing 1101113 ops/sec 4318 MB/sec

[23:57:18] <LouisT> Hello, I seem to be having a memory leak with the node-mongodb-native module, I was just wondering if anyone here would be willing to help me out with it? I'm not even sure if it's considered a memory leak, but it fills uses over 2.5GB of memory when I run my project.

Log file Viewer

Help | Karma | Search:

#mongodb logs for Wednesday the 19th of September, 2012