Log file Viewer
#mongodb logs for Sunday the 15th of July, 2012
[00:31:14] <sessy> When using non ascii charaters on osx 10.6 backspace will REPLACE the non ascii chars with a ?, or delete 2 chars if deleting befre the non ascii char.
[00:37:48] <deoxxa> sessy: what it's doing it deleting one byte, which leaves an invalid utf8 sequence, which displays as a "?"
[00:40:20] <deoxxa> makes me wonder why they don't just use readline or something that works properly
[01:34:40] <jmar777> we're running a replica set, with very low volume, but consistent writes (~10/minute). all of the sudden, queries have begun returning with very high latency (~20 seconds, whereas before they were just a couple ms)
[01:35:12] <jmar777> the only lead I think I've found so far is that mongotop reports local.oplog.rs at 1000ms for both read and write, consistently
[01:35:34] <jmar777> any pointers on where to dig next? not finding much on google regarding really high oplog read/write times
[10:44:38] <DigitalKiwi> if storing stuff with gridfs to be served with the nginx module, is it better to generate a unique filename or is it alright to use the id?
[10:46:48] <DigitalKiwi> i know with like incremental IDs such as in postgres they say not to use the id as it says too much about your data, but i don't know if that stance applies to mongo ids too
[10:55:01] <DigitalKiwi> one more for now, when you link to other document ids, and then that document might be deleted, what's the right thing to do? which I guess is probably pretty app specific...what are some options?
[10:55:25] <mids> DigitalKiwi: http://www.mongodb.org/display/DOCS/Object+IDs#ObjectIDs-BSONObjectIDSpecification
[11:13:37] <kali> DigitalKiwi: it does work. but you need to make all the queries aware of the flag in the app, and the indexes too
[11:17:22] <DigitalKiwi> would as far as the indexes go it pretty much just be a matter of adding the deleted flag to whatever indexes you already have?
[12:51:13] <jondot> is there any way to do findAndModify over a collection (and also get that collection back) ?
[16:25:49] <ahri> hi, i'm pretty new to mongo but i'm trying to use it for a project (to learn these weird ways ;)) and i don't really understand how to get what i would with SQL with a max() -- https://gist.github.com/3117632 is my collection of node documents and i want to pull out max value i have for the "paths.depth" property. can anyone point me in the right direction?
[16:27:18] <ahri> i had the idea to map/reduce them but got tripped up because "paths" is a list, and then i found out about aggregation and simply got confused!
[16:28:26] <ahri> with the map/reduce approach i guess i'd have to do two passes; one to get the max depth per document, then another to get max depth for the collection?
[17:18:16] <Derick> dob_: only with a regular expression search - alternatively, you could store a lowercase variant of the text in a seperate field and search against that
[17:59:56] <wakawaka> I swapped from 1 512 GB SSD to a raid-0 with 512+256 GB SSD. And now the read-performance in MongoDB is kinda sucky :/ .
[18:00:10] <wakawaka> The Production Notes page says "RAID-0 will provide good write performance but provides limited availability, and reduced performance on reads. It is not suggested.".
[18:23:29] <wakawaka> At the moment mongo is under heavy read and modest write. Writes ~2Mb/S Reads ~300Mb/S
[18:24:43] <wakawaka> Lots and lots of querys > 100ms where we hit a index. Querying with explain revelas that index is used both for finding and sorting.
[18:30:47] <dob_> how does a regex find work to get all entries of a collection with ignoring case when sorting?
[19:03:29] <brillopad> Hi guys - I've got a collection that has four elements that I'll want to index. One of them is the date, and just ONE of the other indexes will be used in conjunction with the date per search. Is it best to create three compound indexes with each of these elements and the date together? Or should I do each one seperately?
[19:16:02] <UnSleep> its in mongolab i think its only posible to use mongo correctly in 64bits, isnt it?
[19:16:44] <brillopad> UnSleep - you can use mongo in 32-bit mode, but there's a file size limit of 2GB
[19:17:46] <UnSleep> the databade is now 72mb with a limited space of 240mb. i think i must be doing something wrong
[19:23:08] <brillopad> Couple of things that might add to the size: journal entries - have you added and deleted a ton of documents?
[19:23:48] <brillopad> That should remove any documents that have been deleted but are still in the journal.
[19:25:33] <UnSleep> "message": [ "The repairDatabase command was issued, but the request timed out.", "Usually this means the command was successfully started in the database,", "but it will take a while to complete." ]
[19:26:33] <UnSleep> i only was trying to reate a 10 million documents to play with the index and "relations"
[19:58:59] <vsmatck> How do I drop a collection named "user:session" in the mongo shell? It doesn't like the ":".
[20:05:21] <vsmatck> I can just drop database and use underscore instead. Figured I'd try to learn how to deal with this for future reference.
[21:17:33] <dob_> using regex on find will not handle the case. I tried db.collection.find({name: /./i}).sort({name:1})