Log file Viewer

[01:00:50] <DigitalKiwi> would using gridfs (possibly with the nginx-gridfs) for a very much picture oriented site be a bad idea or would I be better with some cdn or?

[04:11:56] <cagedwisdom> DigitalKiwi, if you are serving static content, a cdn type solution is way better. use mongo for the urls

[04:13:36] <DigitalKiwi> hmm

[04:16:32] <DigitalKiwi> when is gridfs useful? For like videos or?

[04:28:39] <cagedwisdom> DigitalKiwi, I imagine the use case is for files that are changing and need more than one server

[04:30:16] <cagedwisdom> even that can be achieved other ways. if i was going to do something with distributed filesystem i'd be looking at hekafs

[04:39:42] <Diddy403> sup guys

[04:41:41] <Diddy403> anyone care to give me the skinny on why I should use a noSQL database instead of a SQL Server DB?

[04:42:55] <DigitalKiwi> cagedwisdom: when you say static, how static do you mean? Like, there are sites that have lots of images but they are always the same image, that i consider static. but there are sites with user submitted images, so the site is different every time etc.. While each image is static, I'm not sure that is the same as the former case (and thus maybe not as good to use a CDN, or maybe it still is)

[04:44:00] <DigitalKiwi> Diddy403: http://www.mongodb.org/display/DOCS/Use+Cases

[04:46:01] <Diddy403> Is there any way to use Mongo if I'm building in VS2010 rather than like Eclipse?

[04:47:30] <cagedwisdom> DigitalKiwi, in any distributed system it's how often the data changes that affects everything. If the users are uploading new images you don't want to change the image in almost all cases, you want to just add a new image and change the url to point somewhere different.

[04:48:39] <cagedwisdom> DigitalKiwi, the exception would be some huge image where the delta is a small fraction of the image's size.

[04:52:56] <DigitalKiwi> I think we're still not on the same page :(

[04:53:35] <cagedwisdom> DigitalKiwi, what part dont you get/disagree with?

[04:53:58] <cagedwisdom> DigitalKiwi, why dont you describe your use case instead of being so general?

[04:54:03] <DigitalKiwi> :)

[04:54:14] <DigitalKiwi> think something like an image board site

[04:55:22] <DigitalKiwi> where users submit images, maybe a few maybe a lot per user, and they stay there for a long time until they're deleted, people comment on them etc.

[04:55:39] <DigitalKiwi> but they don't replace them just constantly adding more and more

[04:56:19] <DigitalKiwi> Diddy403: sure, mongo is a database it's not tied to any IDE...

[04:56:34] <cagedwisdom> ok so whats the problem? cdn will get you better performance and be cheaper to run then gridfs imo.

[04:56:58] <delhivery> What is EXIT_ABRUBT in mongo?

[04:57:02] <cagedwisdom> cdn is just outsourced master / slave.

[04:58:45] <DigitalKiwi> note that i wasn't saying you were wrong, just wasn't sure if i'd given enough information to give the question i intended :)

[05:08:49] <delhivery> Can someone please help me figure out why mongo db is randomly dying with exit status 14?

[06:26:30] <linsys> delhivery; what does your log say

[06:34:03] <delhivery> linsys, that mongo db exited with status 14

[06:54:07] <trupheenix> hi. i want to know what is the equivalent of stored procs in MongoDb?

[06:59:25] <trupheenix> ?

[07:11:26] <mids> trupheenix: http://www.mongodb.org/display/DOCS/Server-side+Code+Execution

[07:19:28] <trupheenix> mids looks like it's better not to store stuff on mongo db itself. eval takes a write lock for itself? that's not good if you are trying async stuff.

[07:20:57] <trupheenix> mids, can i store my code in a .js file and then ask mongo db to execute it? how does this happen from pymongo for instance?

[10:13:29] <bid> hi, is it a good idea to use mongodb as my "member directory" i mean my users database?

[10:13:47] <NodeX> I do

[10:15:16] <bid> Nodex, can you tell how many users you have? i mean what is the max limit it can support (without sharding)

[10:16:30] <NodeX> 5 million on one server

[10:16:34] <NodeX> server/site

[10:16:44] <NodeX> few hundred thousand on another

[10:17:36] <bid> thanks, any tips for bigunners?

[10:17:51] <NodeX> make sure you index properly

[10:18:26] <bid> sure

[10:19:51] <NodeX> what sort of site is it for?

[10:22:05] <bid> it's more for my mobile app, it will have user login and more info...

[10:30:01] <NodeX> strangely I am just writing a mobile app for one of my sites that uses mongo as it's user store

[10:31:41] <bid> :)

[10:37:04] <NodeX> :)

[10:38:52] <NodeX> trade what?>

[10:39:05] <deoxxa> wat

[10:39:21] <NodeX> bid ?

[10:50:22] <NodeX> :/

[11:18:54] <emilsedgh> ahm, guys, im having a trouble with a query. please take a look at http://paste.kde.org/517508

[11:19:26] <emilsedgh> as you can see, there is a record with name_en = 'Alireza Ghomayshi'

[11:19:43] <emilsedgh> however, when i try to find is using $or, it does not return anything

[11:20:30] <emilsedgh> am i using $or in an invalid way?

[11:22:24] <emilsedgh> ah crap, my version is too old. $or is 1.6+

[11:46:44] <quansai> emilsedgh: Try it without the semicolon at the end? idk

[11:47:00] <quansai> emilsedgh: The command looks fine.

[11:48:55] <quansai> emilsedgh: ?

[12:06:07] <Bartzy> Hi

[12:06:10] <Bartzy> Anyone here?

[12:06:21] <Bartzy> I'm wondering if I should setup LVM in order to use backups, or just use mongodump ?

[12:06:45] <Bartzy> mongodump is a hot backup as far as I know - any reason not to just use it instead of LVM snapshots, which can degrade IO performance?

[12:29:29] <Bartzy> Anyone?

[12:32:48] <quansai> I don't think anyone's up. :P

[14:15:32] <mephju> Hi guys. I have schema like that: { name: Author name", posts : [...] } In my db are several such objects and now I want to retrieve all posts. How can I do that?

[14:15:51] <mephju> Is it even possible?

[14:16:17] <Derick> only as part of each document

[14:16:49] <mephju> Derick, thx

[14:17:20] <mephju> Derick, I assumed that because I couldn't find it in the docs. do u know why it's not implemented?

[14:17:31] <mephju> seems pretty basic

[14:18:07] <Derick> what do you mean?

[14:18:16] <Derick> MongoDB by design only returns documents

[14:18:29] <mephju> it must be an often desired feature.

[14:19:13] <mephju> I would like to retrieve all posts...which are docs

[14:19:46] <Derick> find() will do?

[14:20:03] <mephju> but I understood you that it would return [{posts:[]}, {posts:[]} ]

[14:20:08] <Derick> yes

[14:20:38] <mephju> I want a union of posts

[14:21:01] <Derick> you can't do that by design

[14:21:12] <Derick> unless you use map-reduce, which I wouldn't advice you do

[14:21:22] <Derick> it's a lot simpler to deal with it in the client

[14:21:44] <mephju> it's not done to maintain scalability probably, right

[14:22:30] <Derick> the other option is to redesign your scheme so that each post is a document, with the author an extra field

[14:22:43] <mephju> I had this before

[14:22:51] <mephju> wanted to be more document oriented

[14:23:16] <mephju> sometimes I want to show all posts of an author and I thought this is the way to go in mongo...

[14:23:16] <Derick> both of them are document oriented :P

[14:23:27] <mephju> yeah but the other oneis pretty relational

[14:23:32] <mephju> the one u suggested

[14:23:37] <Derick> you design your schema depending on insert and query patterns...

[14:23:59] <mephju> yes. but I just assumed the union would be possible

[15:04:35] <Bartzy> Is there any reason to over provision current SSD drives ?

[15:04:45] <Bartzy> Derick: Perhaps you know? ^ :)

[18:59:44] <hadees> is it possible to update a field on all the documents in a collection using a javascript function that calculates the value from the document?

[19:00:27] <hadees> I want to implement this ranking function in mongodb https://gist.github.com/2636355 but i was thinking instead of having ruby make a ton of queries I could just make one?

[19:06:57] <mids> hadees: you can do some server side execution; but that imposes a global write lock on the collection

[19:07:04] <mids> is this a one-time task?

[19:12:05] <hadees> miss, nope it needs to be updated regularly

[19:12:13] <hadees> mids:

[19:13:04] <mids> only when someone votes right?

[19:29:46] <hadees> mids: nope because time affects it too, i'll prob do it on a cron job

[19:43:30] <mids> hadees: isnt 'date' the time the story was submitted?

[19:45:39] <hadees> miss, i guess that is true, i've been playing around with a few different hot algrothims, i think you're right about the reedit one. So what do you suggest?

[19:46:25] <mids> I suggest that the 'hot' score only needs to be modified if someone votes on it

[19:47:10] <mids> then if you put an index on that attribute; you dont need to run cron jobs or anything

[20:15:58] <bingomanatee> how large can a Mongo record get before it degrades performance? (use case: blog articles)

[20:16:50] <tmw> win 43

[20:17:43] <Derick> bingomanatee: difficult to say really

[20:17:47] <Derick> tmw: that's a lot of windows!

[20:26:31] <weeb1e> Hello

[20:27:23] <weeb1e> How can I unset a key in a document? update with $unset: {key: 1} does not seem to work, the key remains as an empty array

[20:33:09] <weeb1e> Ah I was missing multi :)

[20:36:40] <mids> :)

[21:14:21] <diogotozzi> Anyone here was at MongoSP?

[21:16:05] <Derick> sadly not :-/ I'm sure it isn't pouring down there all summer :-)

[21:19:22] <diogotozzi> lol

[21:22:23] <diogotozzi> Derick: Did you create the pecl VLD?

[21:23:48] <Derick> yes

[21:24:14] <diogotozzi> Derick: Wow, nice!

[21:24:51] <Derick> xdebug too ;-)

[21:25:16] <diogotozzi> Derick: Man, you are my idol! hehe

[21:26:49] <diogotozzi> Derick: Yesterday I met a lot of guys from 10gen at MongoSP. You should have come

[21:29:09] <Derick> it's a bit far and I do plenty of traveling already

[21:29:48] <diogotozzi> Derick: that's true =/

[21:36:12] <DiogoTozzi> Derick: I gotta go, we talk more later. See ya

[22:58:41] <Peste_Bubonica> Hi all!