Log file Viewer
#mongodb logs for Wednesday the 29th of May, 2013
[05:23:13] <belak> I'm looking at using nosql for an app I'm writing... and I'm trying to determine how I'd organize the data I need. Essentially, there are a number of posts and each post will have a file associated with it, as well as a bunch of tags. All that's relatively simple, nesting the tags inside some sort of posts object... but how would I find all posts with a specific tag? Iterating through everything seems really efficient, but please correct me if I'm wr
[05:36:33] <adoniscik> what is the idiomatic way to filter elements of a collection by co-occurrence? Say I have collection of tuples with (document, user, rating). For a given pair of users, how can I obtain the relevant tuples with *common* documents?
[06:04:52] <sureshsaggar> Is there an API in mongodb that returns the delta of updated documents after a certain timestamp?
[06:05:55] <sureshsaggar> Just like secondary replica set getting updated after a certain interval, we need to store the same delta updates some place else programitically
[07:19:08] <groundup> I am getting assertion: 13106 nextSafe(): { $err: "Invalid ns [.kidtrips]", code: 16256 } when I try to restore mongorestore --collection kidtrips /kidtrips/mongobackup
[08:18:02] <ncls> Nodex I was browsing an archive of this channel when I saw you were talking about Cahors with someone, are you from France ?
[09:06:09] <johmue> i have a weird behaviour with chrome and mongodb under Win7, starting Chrome causes a boost in perfomance when using the aggregation framework and i don't know why
[09:06:59] <johmue> no other program causes this behaviour - it runs 6-8 times faster with chrome opened than running stand alone
[09:22:14] <Number6> johmue: Both MongoDB and Chrome use the V8 javascript engine... I'd imagine (and I can only say this for Linux, not Windows), that the OS is caching the V8 binary in RAM as multiple processes are using it. In Linux, the OS would have cached the binary in RAM anyway
[09:27:06] <Kosch> Question related to replication: I've a hidden instance which appears only from time to time and has changing IP. Is it possible, regardless if this is listed in the members, to sync it?
[09:28:18] <kali> Number6: only if they're using it as a shared library (system wide .dll or .so) and i doubt very much they do
[09:29:49] <kali> johmue: this is really weird. the AF does not make use of V8 anyway, so i really think that's not it
[09:35:15] <johmue> i took a closer look at the code noticed I was wrong - the aggregation framework was not involved... it is a normal query that runs through the whole dataset. So I have to check if it is also the case for the aggregation framework - i'm sorry
[09:35:22] <hjrnunes> Hi everyone, is it possible to use the cursor.min / max with pymongo? I'm trying to get around https://jira.mongodb.org/browse/SERVER-8135 but I can't use the array solution as I'm already indexing an array for that query
[10:33:19] <sorre> hi, i am trying to use a $each modifier in php but i cant find any examples, is this the right usage? $this->dbUser->update(array('_id' => new \MongoId($idUser)), array('$push' => array('follows'=> $aQolles)), array('each' => true));
[10:52:35] <hjrnunes> Hi everyone, is it possible to use the cursor.min / max with pymongo? I'm trying to get around https://jira.mongodb.org/browse/SERVER-8135 but I can't use the array solution as I'm already indexing an array for that query
[11:57:07] <zapzup> which finds an image, and creates a DBREF to that image inside another document (via update)
[11:58:01] <zapzup> the script doesn´t work, becaus the image._id syntax doesn´t seem to give the value. it seems like the variable isn´t set at all. can someone help? (im absolutely new to mongodb)
[13:12:40] <remonvv> "nscanned:0 nupdated:1 upsert:1 keyUpdates:0 locks(micros) w:169178 168ms", anyone any idea what the "w" is here?
[13:21:03] <Nodex> trying to tail my log to see what it might be but I don't have any long running queries :(
[13:43:48] <Nodex> remonvv : "w" = the length of time in microseconds that MongoDB waited for a lock
[14:19:10] <arthurnn> hi.. i jsut finished draining a node on a shard cluster... but i still see deletes on those boxes after the draining is done.. any reason for it??
[14:19:35] <Foxandxss> Hi, Im going to start into NoSQL world and I see in a lot of places that there is no join, I guess, no more relationships. Do you know any document that explain how NoSQL works or compare to relational dbs, thanks :P
[14:23:31] <Nodex> and you shuld always remember the path of least resistance will be the fastest which means embedding (mostly) makes more sense than relating
[14:25:21] <Foxandxss> I come from an easy of ORM "Ask me, I give you all you want, well formated and ready for business" :P
[14:29:01] <Foxandxss> there is not a concrete question. I'm just learning node + mongo and things like that and in some sites I see that there is no relations, but moongose seems to have some kind of relations (just, 1-n and 1-1 but not n-1 or n-n), so I just got confused, someone wants to relate, others don't
[14:30:28] <Nodex> Foxandxss : that's down to the driver / ORM creators personal preference I would guess
[14:30:58] <Nodex> some ODM/ORM's support relations by automagically getting the other document for you
[14:32:57] <starfly> Foxandxxx: recommend the free, online MongoDB developer training that 10gen offers (see https://education.10gen.com/)
[14:35:03] <jsec> I'm working on a daemon that interfaces with inotify to add file information into mongo every time a file is added to a specific folder. the parameters for said information are stored in a static GArray. There are 2 separate queries I need to perform on the db to get a couple of validation checks. Problem is, every time I run one of the queries, the mongo_find cursor overwrites the memory address for the static GArray, and I can't figure out if it'
[14:36:20] <starfly> Foxandxss: they are certainly helpful, I've taken the Python and Java courses (and DBA one) and got a lot out of all of them. Good luck and sorry to mangle your name earlier
[14:36:54] <Foxandxss> starfly: liked the pr0n part :P. I will checkout them, the python one seems good, I don't like java at all :P
[14:41:00] <Foxandxss> starfly: that's bad, IRC is REALLY interesting and this server has a lot of good channels, for everything
[14:41:35] <starfly> Foxandxss: cool, just have to harmonize it with the usual never-ending workload, thanks
[14:42:22] <Foxandxss> starfly: do like others, 24/7 connected but none have ever seen then talking :P. Nah, sometimes you can find good talks around there
[14:52:17] <bartzy> In the PHP driver, we now have to do new MongoBinData($blob, MongoBinData::BYTE_ARRAY) ?
[15:09:52] <quetz> How can i get name of replica set? I can connect to it with mongo client, but when i connecting with pymongo, i have "the replicaSet keyword parameter is required." exception
[15:15:46] <double_p> ok.. i've some secondary/backup site. i want to mitigate the point that when this node goes down (or drops connectivity) a reelection/rs.reconfigure happens. strategies?
[15:22:07] <quetz> starfly, so field "_id" = name of replica set? I am getting "No replica set secondary available for query with ReadPreference SECONDARY" error so i am not sure is it because wrong name or server problem..
[15:24:15] <johntron> i have an array of subdocuments with unix timestamps as a field. How do I use the timestamps to add a human-readable date to each of the subdocuments?
[15:25:09] <johntron> i think this is two questions: 1) how to add a field to all subdocuments in an array and 2) how to create a date string from a (timestamp) field.
[15:26:56] <bobinator60> and the error i get is: "attributes.attribute_type" : {"$in", self.md_fields}
[15:39:30] <bartzy> Derick: Are we now supposed to do: new MongoBinData($blob, MongoBinData::BYTE_ARRAY) for binary data ?
[15:40:00] <bartzy> ongoBinData::__construct(): The default value for type will change to 0 in the future. Please pass in '2' explicitly.
[15:44:20] <starfly> quetz: sorry, been away, yes, the _id field in the replica set name. Looks like your read pref is in the way, connect to the primary and run that rs.conf() command or "use local" and "db.system.replset.find().pretty()" instead
[16:04:42] <johntron> For a collection of subdocuments, adds a human-friendly date string using an existing timestamp
[16:07:45] <quetz> starfly, on pretty() command it answer error: { "$err" : "not master and slaveOk=false", "code" : 13435 }
[16:08:29] <quetz> the problem is, admin asks me if my report system can connect to secondary (primary just closed for queries)
[16:22:02] <epirusl1> db.collection.find(something) .if the result is None,and How to get the null value
[16:22:30] <harenson> quetz: http://api.mongodb.org/python/current/examples/high_availability.html#secondary-reads
[16:43:16] <epirus1> I have a stupid question: If a query can't find any document. How do I know it ? no return value ?
[16:58:38] <epirus1> harenson, good ideas . but I have another question: db.collection.find().skip(10).count() the skip seems does't work. I get the same count number
[16:59:44] <harenson> epirus1: db.collection.find().count() == db.collection.find().skip(10).count() ?
[17:01:26] <starfly> epirus1: I believe skip() will affect a returned cursor, but not the count() value
[17:14:19] <harenson> epirus1: try this: foos = db.foo.find().skip(10), then execute: foos.explain()['n']
[17:15:38] <harenson> epirus1: or you could do: print db.foo.find().skip(10).explain()['n'] if you just want the count result
[17:28:46] <phaidros> hi, does a replica set of two make sense? if the first one fails, the second one is no majority, what will happen? (I am not sure if I understand the docs completely here)
[17:31:09] <quetz> pymongo.errors.AutoReconnect: No replica set members available for query with ReadPreference SECONDARY_PREFERRED
[17:40:59] <bobinator60> any mongoengine people on here? is there any reason to use the mongoengine queries instead of using direct mongodb query documents?
[17:42:50] <starfly> quetz: con = MongoReplicaSetClient("10.100.10.44:27017", "10.100.10.4:27017", replicaSet='pwsocagr', read_preference=ReadPreference.SECONDARY)
[17:47:25] <quetz> so maybe i need to have access to this IPs too? admins only opened firewall on b519 for my IP
[17:48:23] <starfly> quetz: what's the output from nslookup b519.nivalnetwork.com? Also, can you upgrade your MongoDB, looks like you're running 2.0.1
[17:52:58] <starfly> quetz: so, this is a network access issue, I hadn't realized that you were running your Pymongo outside a fire-walled environment that was using NAT. You'll presumably need a public IP address for the secondary box and use that in your connection setup
[18:02:57] <starfly> quetz: when you ran the mongo shell command (in Gyazo screenshot), were you outside or inside the firewall? That connected to the secondary at 108.168.173.3 (public b519 IP addr)
[18:12:31] <quetz> ok, trying. btw both commands - connection and iterating through result goes very slowly. its like it waiting 1 min
[18:16:24] <starfly> quetz: so, I'm wondering if the root cause is that you're replicating inside the firewall with NAT'ed IPs and when the Pymongo client goes after the connection, it tries to use the NAT'ed address instead of the public address
[18:17:50] <starfly> quetz: if you control the whole environment and it isn't currently in production use, perhaps you can re-establish the replica set with the two public IPs
[18:18:50] <starfly> quetz: or, perhaps if you have both public IPs or hostnames, you can specify those in the Pymongo connection and it'll work--certainly an easier try.
[18:27:36] <quetz> starfly: you was right, pymongo go through this 10.100.. members and all connections fails -_-
[18:27:47] <starfly> quetz: good luck and BTW, you might want to go with another secondary or at least an arbiter to allow proper election of an available secondary if the primary goes down
[18:28:28] <starfly> quetz: OK, good to know, so you might get the public IP for the primary MongoDB host and put both in your Pymongo connection setup
[18:45:08] <bhangm> hello folks, does anyone know if a straight upgrade from 2.0.9 replset w/ auth to 2.4.3 is possible? The docs do not seem to indicate otherwise, but the last comment on this (https://jira.mongodb.org/browse/SERVER-6897?focusedCommentId=160903&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-160903) JIRA issue mentions upgrading via 2.2.1
[19:13:03] <acalbaza> does anyone have a good auditing solution? i have a collection that i'd like to track changes for. simply capturing the state of the collection along with a timestamp may be enough for me.
[19:25:30] <oblio> i'm trying to build mongodb 2.4 from source on smartos and i'm getting this error: AssertionError: Unsupported architecture: i86pc:
[19:27:36] <starfly> oblio: what's the underlying hardware, 32 bit-based? Did you get the 32-bit binaries? http://www.mongodb.org/downloads
[19:30:59] <starfly> oblio: the assertion error suggests that 32-bit architecture is being used, so you have to at least match with 32-bit MongoDB binaries. I have no idea if SmartOS has enough to support MongoDB
[19:31:54] <starfly> oblio: going back to your earlier comment, sounds like you're compiling MongoDB source.
[19:38:22] <starfly> oblio: i86pc spec includes 32 and 64 bit (x86 and x86-64), since your SmartOS is 64-bit, sounds like the 64-bit switch should do the trick
[19:48:03] <starfly> quetz: not sure if you're still out there, but although you can't upgrade your MongoDB server side infrastructure, you should try the latest Pymongo (if not already using it). With the mongo shell navigating the same network topology, you'd think Pymongo could toe the line for that as well.
[20:01:11] <jcalvinowens> Quick (hopefully not stupid) question: is it possible to negate $match in an aggregation query? As in {"$match": "field_name": {"$not": "exclude_me"}}? (That produces an error)
[20:34:16] <johntron> i just figured out how to use third-party javascript libraries with mongo and added a few years to my life :D
[21:19:33] <robovoyo> Does anyone have any insight into why sorting by reverse natural order would be slow (on a non-capped collection)? I have a query that returns in ~200ms on average. When I sort by _id descending it's about 20 times slower. When I sort by $natural descending, it's about 10 times slower.