pmxbot IRC Log Viewer

[00:15:07] <prophit> If I have documents with variable numbers of keys, how can I get a list of the keys on the document? or a count of the number of keys?

[04:08:57] <prophit> if I have a collection with docs like {id: 'somekey', count: 4}, how can I use the aggregation framework to compute the average count?

[08:39:01] <[AD]Turbo> hola

[08:42:13] <ethanol> how can I control how many child processes mongodb parent spawns?

[08:42:20] <ethanol> I couldn't find anything about this in config documentation

[08:42:40] <ethanol> it spawns like 9 of them, whilst I could probably do with 1 or 2 :/

[14:41:57] <lfamorim> Hello!

[14:42:07] <lfamorim> How could I change a DBRef collection?

[14:42:19] <lfamorim> just the collection

[15:37:07] <andreligne> Hi! I have a question regarding using mongodb with nodejs :)

[15:38:09] <NodeX> better to just ask than ask to ask

[15:39:01] <andreligne> True

[15:39:33] <andreligne> I'm wondering if it's a good idea to have all the database-requests in one "controller"-file and then use this within all other controllers

[16:04:19] <ethanol> can someone point me at where in the documentation it describes how to reduce the number of child processes mongodb spawns at startup?

[16:34:04] <flock3> Hey guys, I have some data in this format: http://pastie.org/6064047 I think I need to do $pushAll to append the event_alerts key to something like this: http://pastie.org/6064154 - how do I format this?

[16:50:12] <NodeX> $pushAll :{599 : Object}

[16:50:23] <NodeX> $pushAll :{"599" : Object}

[16:50:49] <Killerguy> hi, does the nagios plugin to monitor mongo cluster works well?

[16:50:57] <Killerguy> does someone got any advise about it?

[16:53:37] <tom0815> hi killerguy, I used the mongodb plugins for a few month and they are doing their work

[16:54:44] <owen1> indexes questions - i have a find by user_id and app_id. does it mean i should have compound index?

[16:55:22] <tom0815> of cause most work is to adjust the params of the nagios plugins to match your setup

[16:55:22] <NodeX> yes

[16:55:28] <vlad-paiu> Hello. have some weird issue with mongodb.. in the CLI I do something like db.col.insert({ "version": 4, "name": "address"}) , and then I fetch the record from with the C driver, but I get the 4 value as a BSON_DOUBLE instead of a BSON_INT ... is that expecting, or am I inserting it wrong..

[16:55:38] <NodeX> compound index on user_id, app_id

[16:55:49] <owen1> NodeX: i also have an upsert based on app_id and device_id. what does that mean in regards to indexes?

[16:56:12] <kali> vlad-paiu: you need to make the types explicit in the shell, as js is not typed

[16:56:22] <NodeX> does app_id

[16:56:32] <NodeX> + device_id get used for anything else?

[16:56:44] <kali> vlad-paiu: http://docs.mongodb.org/manual/core/shell-types/

[16:57:04] <owen1> NodeX: let me see

[16:57:26] <NodeX> you'll want an extra index for it unless it's not an important upsert

[16:59:02] <owen1> NodeX: i don't really use the app_id + device_id for anything but that upsert. the main thing is the find (user+app)

[16:59:27] <Killerguy> ok tom0815 thx

[16:59:28] <NodeX> owen1 : it will depend on what you're app tollerance is like for a slow upsert

[16:59:37] <Killerguy> with a cluster about 100shard it will works?

[16:59:42] <vlad-paiu> kali: ok, got you. thanks, worked

[16:59:43] <NodeX> personaly I would add the index

[17:00:25] <owen1> NodeX: so user+app is %100 needed, and maybe also device?

[17:00:48] <NodeX> device wont get to use it

[17:01:29] <owen1> NodeX: when u said "I would add the index". what indexes did u meant?

[17:04:36] <NodeX> both

[17:05:25] <owen1> NodeX: ok. user+app but no need for user+app+devices. thanks!

[17:07:26] <NodeX> user+app+device will never get used

[17:26:38] <alexwaters> does anyone know how I can remove just one document with pymongo? can i do monDB['collection'].remove({'orderid':'123'}, True, Safe=True)

[17:27:48] <alexwaters> also, the docs says remove safe is deprecated and I should use w. Anyone have docs on how I should be checking that critical write/remove commands were successfully processed?

[18:00:29] <vlad-paiu> Hi. Another C question, if anybody's around.. is there an easy way to determine how many 'rows' a cursor has in the mongoDB C library ? Is it safe to save the initial cursor, then iterate with mongo_cursor_next() until no more rows left, and then start from the beggining with the initial cursor ? or will this mess something around in the mongo C library internals ? :)

[18:07:50] <vlad-paiu> tried above, and didn't work. not sure if it's the library's intended way of working, but did m_cursor->reply->fields.num which seems to return the correct value

[18:24:37] <Killerguy> how can I know if my balancer is working too much?

[18:24:57] <Killerguy> I mean, I want to know if my shard key is good enought to shard properly

[18:27:09] <jmpf> I have a secondary that is *way* behind a primary but rs.status() indicates otherwise - wanting to run repairDatabase on the secondary, then demote primary and do the same - is that going to be a problem?

[19:18:23] <nb-ben> using the JavaScript driver, calling collection.find( ... ) does not require a callback? or does the toArray( function() {}) call actually perform the query and calls its callback after fetching the results ?

[19:19:17] <beandip> Anyone here using the edelight cookbook for mongodb?

[19:20:00] <beandip> I'm running into errors getting it to load in a vagrant/chef-solo wondering if anyone here has this running that I can bounce some questions off

[19:28:43] <tubbo> hey guys, does Mongo lock when you write to the DB?

[19:28:48] <tubbo> (like *SQL stuffs?)

[19:29:19] <ron> Yes.

[19:43:47] <tubbo> ok

[19:47:30] <Garito> hi, all!

[19:48:49] <Garito> may I ask about aggregation framework here?

[19:50:39] <Garito> hello?

[19:50:49] <Garito> 358 users and nobody say anything

[19:50:51] <Garito> ?

[19:51:02] <mr_smith> don't ask to ask. just ask.

[19:51:08] <Garito> sure

[19:51:23] <Garito> need to count when a field ends with a concrete substring

[19:51:30] <Garito> how can i do that?

[19:51:43] <Garito> I have this

[19:51:44] <Garito> pyExpr:{

[19:51:44] <Garito> '$match': {

[19:51:44] <Garito> 'time': {

[19:51:44] <Garito> '$gte': utilsFecha['esFecha']('%sT00:00:00Z' % path('Yanged/yQueryString/Inicio/0')),

[19:51:44] <Garito> '$lt' : utilsFecha['esFecha']('%sT23:59:59Z' % path('Yanged/yQueryString/Fin/0'))

[19:52:10] <Garito> it works ok

[19:52:11] <mr_smith> use a pastebin.

[19:52:19] <Garito> but now I need to add the counting

[19:52:20] <Garito> sorrey

[19:52:22] <Garito> sorry

[19:52:43] <Garito> http://pastebin.com/PDarGg7n

[19:52:53] <Garito> I have this aggregation project

[19:53:04] <Garito> who count me the number of visits

[19:53:13] <Garito> with the ones with authenticated users

[19:53:29] <Garito> now I would like to add when the visit ends with a substring

[19:53:33] <mr_smith> i would think your $group was counting correctly.

[19:53:33] <Garito> in the same project

[19:53:43] <Garito> yes

[19:53:45] <Garito> it works ok

[19:53:51] <Garito> what I need is to add

[19:53:54] <Garito> another counter

[19:54:07] <Garito> to count the visists who ends with a substring

[19:54:18] <Garito> but I can't find the correct one

[19:54:23] <mr_smith> not sure how to do that. sounds like a separate aggregation.

[19:54:32] <Garito> yeah

[19:54:38] <Garito> I could do in this way

[19:54:51] <mr_smith> but if that's what you're doing, you could use a $match with a regex.

[19:54:53] <Garito> but then I need to find 2 times the same information

[19:55:08] <Garito> in a different project I could

[19:55:25] <Garito> but i don't want to visit the whole table again

[19:55:33] <Garito> for making another counting

[19:55:42] <Garito> sounds very bad

[19:56:02] <mr_smith> how often are you running these queries?

[19:56:08] <Garito> on demand

[19:56:31] <Garito> will be part of an ecommerce dashboard

[19:56:40] <Garito> want to count:

[19:56:40] <Garito> visits

[19:56:48] <Garito> authenticated users from this visits

[19:56:55] <Garito> checkouts

[19:57:01] <Garito> etc

[19:57:24] <NodeX> what

[19:57:25] <NodeX> are

[19:57:27] <NodeX> you

[19:57:29] <NodeX> trying to

[19:57:30] <NodeX> do

[19:57:43] <Garito> an ecommerce dashboard

[19:57:46] <mr_smith> don't encourage him.

[19:58:33] <Garito> so?

[19:58:37] <Garito> what do you think?

[19:58:50] <NodeX> think a out what?

[19:58:54] <NodeX> aboout*

[19:59:43] <andybons> would someone mind explaining to me why an aggregation result set's length would be different than the result once its been run through { $group: { count: $sum: 1}} ?

[19:59:57] <Garito> I think in a conditional with a regex

[20:00:05] <Garito> will this be possible?

[20:00:11] <mr_smith> i think you have a number of options, but i'm pretty sure you're going to have to do multiple aggregations. if you're worried about load, you can query secondaries or cron the queries every 5 minutes and cache the results.

[20:00:29] <Garito> shit!

[20:00:44] <Garito> I was thinking that aggregation make this posible

[20:00:55] <Garito> am I correct?

[20:01:41] <Garito> If not aggregation isn't useful

[20:01:46] <mr_smith> so you want to do multiple matches with independent counts in a single aggregation pipeline?

[20:01:58] <Garito> I'm doing it right now

[20:02:07] <mr_smith> then why are you asking?

[20:02:12] <Garito> with visits vs authenticated users

[20:02:22] <Garito> because I need another count with a reges

[20:02:25] <Garito> regex

[20:02:27] <NodeX> personaly I aggregate every X mins and reduce down into different collections... ths is the best way to d near realtime aggregation

[20:02:41] <mr_smith> and i'm +1 with NodeX

[20:03:19] <Garito> so mongo does not make this aggregations?

[20:03:36] <NodeX> mine?

[20:03:37] <NodeX> yes

[20:03:46] <Garito> at least on demand

[20:03:54] <NodeX> on demand is a bad idea

[20:04:15] <Garito> fuck any other database make this possible

[20:04:26] <NodeX> then use another database ;)

[20:05:40] <Garito> really?

[20:07:56] <NodeX> why are you using mongo for this?

[20:08:17] <Garito> I like nosql schemaless

[20:08:32] <NodeX> then you'll have to work in it's bounds

[20:09:01] <Garito> don't seems to me so big deal what i need

[20:09:13] <Garito> seems to me a normal operation

[20:09:40] <NodeX> realtime aggregations?

[20:09:49] <Garito> yeah

[20:10:08] <Garito> counting visits

[20:10:13] <NodeX> upto about 50k rows on a 16gb ram quad core it's fine

[20:10:25] <NodeX> s/rows/documents

[20:11:24] <mr_smith> unless you're amazon, i can't imagine performing multiple aggregations is a problem.

[20:11:37] <Garito> oh man!

[20:11:52] <Garito> I need to count in the same time period several things

[20:11:52] <NodeX> the locks end up being the proble

[20:11:57] <NodeX> +m

[20:12:12] <Garito> and you aks me to make several aggregations

[20:12:27] <Garito> same time period is the key

[20:12:32] <NodeX> you asked us

[20:12:38] <mr_smith> NodeX: i thought only writes lockd.

[20:12:45] <Garito> need to read again and again the same info to aggregate different things

[20:12:51] <Garito> sounds crazy for me

[20:13:05] <NodeX> database level locking

[20:14:20] <mr_smith> NodeX: okay i'm with you. could just hit a secondary.

[20:14:53] <NodeX> no point, just do what I originaly said and aggregate every X mins

[20:15:20] <Garito> ok

[20:15:21] <mr_smith> i'm fine with that too and that's how i'd solve it.

[20:15:30] <Garito> end way there

[20:15:36] <Garito> in my opinion

[20:15:44] <Garito> I wonder what you said

[20:15:49] <Garito> but i have to accept

[20:15:51] <Garito> it

[20:15:56] <NodeX> I dont

[20:15:58] <NodeX> understand

[20:16:00] <NodeX> a

[20:16:03] <NodeX> word you justy

[20:16:05] <NodeX> said

[20:16:21] <Garito> thanks

[20:16:25] <Garito> and see ya

[20:16:34] <NodeX> good luck :)

[20:16:54] <mr_smith> phew

[20:17:41] <kali> :)

[20:17:47] <mr_smith> yeah, but how else are you going to get WEBSCALE!

[20:18:13] <NodeX> rdbms without joins

[20:18:20] <NodeX> and a large cache

[20:18:30] <mr_smith> that doesn't improve anyone's resume.

[20:28:18] <NodeX> +1

[21:01:15] <owen1> i want to add a compound index - app_id and user_id. i only have 3 types of apps. does it mean the user_id should be the first index and the app_id second?

[21:01:27] <saml> how can I saerch for documents that has a sub document (nest level not known) with a key val header_size: 2 ?

[21:01:33] <owen1> (i'll have many user ids)

[22:14:45] <sfa> has anybody used the "mongodb" driver in Node.JS?

[22:15:39] <someprimetime1> aka used mongodb with node.js?

[22:15:47] <sfa> yup

[22:16:23] <sfa> I'm having trouble with doing an "npm install mongodb" ... It is complaining about "pre" versions cannot be installed

[22:16:29] <someprimetime1> sudo

[22:16:37] <someprimetime1> wait

[22:16:42] <someprimetime1> are the requests just not going through?

[22:16:46] <someprimetime1> paste the error

[22:16:52] <sfa> k one sec

[22:17:11] <someprimetime1> normally I `sudo npm install mongodb`

[22:17:35] <sfa> http://pastebin.com/kreSdDtz

[22:18:22] <sfa> let me try sudo

[22:18:38] <sfa> same thing with sudo

[22:18:39] <someprimetime1> yeah

[22:18:44] <someprimetime1> that's only if it won't GET it

[22:18:51] <someprimetime1> hm looks like you have a previous version it's conflicting with maybe?

[22:19:01] <someprimetime1> what happens when you type `mongod` in the cmd prompt

[22:19:17] <someprimetime1> are you on a linux machine?

[22:19:27] <sfa> yes it's a linux machine

[22:19:44] <sfa> and this machine doesn't have mongo installed on it.

[22:21:59] <someprimetime1> try asking in #node.js

[22:22:05] <someprimetime1> I'm not sure tbh

[22:22:19] <sfa> okay thanks for helping someprimetime1

[22:22:21] <sfa> :)

[23:19:48] <solidus-lake> whoa, mongod doesnt have a —config option anymore?

[23:19:49] <solidus-lake> what the hell

[23:19:55] <solidus-lake> so how are we supposed to deploy different configs

[23:20:05] <solidus-lake> a long param list manually in the init file :?

[23:38:58] <solidus-lake> hmm

[23:39:01] <solidus-lake> so i started my master

[23:39:09] <solidus-lake> i started the slave and told it of the master

[23:39:12] <solidus-lake> how do i start an arbiter

[23:40:47] <owen1> i want to add a compound index - app_id and user_id. i only have 3 different apps ids. does it mean the user_id should be the first index and the app_id second?

Log file Viewer

Help | Karma | Search:

#mongodb logs for Tuesday the 5th of February, 2013