Log file Viewer
#mongodb logs for Monday the 14th of July, 2014
[07:27:44] <VeeWee> Mongod won't start. I tried repairing, removing lock file, killing old process and removing pid file. No logs are written. Any ideas?
[07:34:40] <VeeWee> switched them to mongod user, but they got switched back to root for some reason
[07:41:24] <yruss972> Hi, I'm trying to understand the serverStatus output. Is it normal for recordStats.local.accessesNotInMemory to be as high as 2666252 ?
[08:00:13] <rspijker> so, basically. That number for local.accessesNotInMemory is the amount of times mongod had to fetch a page that wasn’t in memory for the local DB
[08:00:56] <rspijker> that’s not all that strange. The local DB is written to every time you write to the DB (since the oplog lives there)
[08:04:14] <rspijker> There was someone in here the other day as well which reported very large vsize…
[08:05:20] <rspijker> well… those numbers being high (accessesNotInMem) are a logical result of the server running out of resident memory
[08:05:59] <yruss972> I don't understand how I could be running out of memory when all my DBs dump into 105MB of files?
[08:06:05] <rspijker> from what you’re saying the amount of actual data in the system isn’t very large
[08:46:21] <shambat> sflint: I want to use mongodb as a db instead of storeing data in huge json files, not for speed, but to save memory usage. I'm using the data in a python script, and the json files are quite large, taking up a lot of memory
[08:48:07] <rspijker> yruss972: how large is you local db? (By default, mongodump doesn;t dump the local)
[08:49:02] <shambat> so not gigantic, but when they're loaded into mamory, my VPS with 4GBs of memory crashes due to full memory
[08:51:10] <rspijker> yruss972 it can be. It’s a capped collection (well the most important on in there, the oplog, is) and it defaults to 5% fo free FS space on 64bit linux
[08:54:49] <yruss972> sflint: "The oplog (operations log) is a special capped collection that keeps a rolling record of all operations that modify the data stored in your databases." ?
[08:56:32] <shambat> sflint: sorry I was afk there ... right now I have nothing in mongodb. I have a python script that does json.load(myfile), which is very demanding on memory. I was hoping that storing the data in mongodb would allow me to just query the database instead of loading the whole thing to memory, thusly preventing the server to crash due to lack of memory
[08:57:29] <rspijker> var c=0;use admin;db.runCommand({"listDatabases":1}).databases.forEach(function(x){db=db.getSiblingDB(x.name);c+=db.stats().dataSize;})print(c);
[08:57:38] <sflint> if you use an index on the data you will be fine....then the query will be fast as well
[09:00:48] <rspijker> yruss972: try this one instead: var c=0;db.adminCommand({"listDatabases":1}).databases.forEach(function(x){db=db.getSiblingDB(x.name);c+=db.stats().dataSize;});print(c);
[09:01:11] <shambat> sflint: this is a script that runs hourly and will add data and compare stored data
[09:03:27] <yruss972> do you think limiting the oplog size will help- most of that seems to be the local db?
[09:04:34] <sflint> shambat: also if you write your queries correctly and use projection you won't return the entire set of data
[09:13:09] <rspijker> in an active system it is normal and desirable to have almost all RAM utilized
[09:17:24] <yruss972> I restarted some of the other machines in the cluster last week for similar issues (left this one to try to find the problem)
[09:42:56] <JT-EC> yruss972: The issue seems to be that SmartOS zones don't have absolute memory limits and mongo doesn't seem to be able to cope very well with that. SmartOS lets you take more memory but swaps it and it kills performance as you're finding.
[09:43:41] <JT-EC> Because of the soft mem limits and (from the docs) "MongoDB automatically uses all free memory on the machine as its cache" this tends to screw SmartOS users.
[09:45:50] <yruss972> ok- but I still don't understand why it should be trying to take more memory when I only have 105MB of data :?
[09:51:16] <JT-EC> yruss972: Look at what I quoted from the docs - "all free memory" - doesn't mention anything about your data size only mongo's greed.
[09:58:51] <nylon> anyone know why text search data that contains periods have strange behavior on search results?
[10:38:53] <dragoonis_> Looking to effectively do upserts on my collections, is this a popular thing for mongoDB?
[10:39:27] <dragoonis_> I need a composite unique index of "survey_id, category_id, results_date (ISODate)) to upsert upon
[10:49:04] <dragoonis> I had another mongoDB question about indexes, i just made it now - maybe you have some feedback for me here too? :) http://stackoverflow.com/questions/24734811/hiow-to-mongodb-ensure-index-upon-collection-creation
[11:00:06] <dragoonis> > db.company_category_score_months.ensureIndex( { "survey_id" 1, "category_id": 1, "results_date": 1 }, {unique: true} )
[11:07:56] <phr3ak> hello, do I have to use cman/rgmanager if I want HA? currently it is not clear for me how mongod could work in HA?
[11:11:41] <phr3ak> so if i set the database server ip in our application and if the node will be unreachable how to know the application the ip address of the working node? how the HA works?
[11:14:44] <rspijker> if you pass a single mongos, then you’re just moving your single point of failure
[11:16:32] <phr3ak> if i understand well in the application we have to set more than one node in the connection string?
[11:28:14] <phr3ak> i don't know what is that arbiter but someone told to me we want to use this redhat clustering for the arbiter...
[12:33:04] <backSlasher> we are expiriencing really slow queries (>3s) with a collection that holds medium (<10MB) files.
[12:33:04] <backSlasher> I don't have a very high lock percentage (~10%-20%) but I do see a very high write queue on my disk, once every ~20s.
[12:33:04] <backSlasher> AFAIK it's no the data flush because the console says that they are one minute apart. "accessNotInMemory" is lower than 1 for the relevant collection.
[12:35:19] <backSlasher> it's an upsert based on id, and idhack is true for these queries, so index is probably not the issue
[12:40:38] <dragoonis> I'm talking the mongodb query log, how do I identify if the query being ran has hit an index or not ?
[12:41:55] <backSlasher> dragoonis, first enable profiling http://docs.mongodb.org/manual/tutorial/manage-the-database-profiler/
[12:44:10] <backSlasher> dragoonis, http://docs.mongodb.org/manual/reference/method/cursor.explain/#cursor.explain
[12:46:52] <backSlasher> dragoonis, http://blog.mongohq.com/explain-explain-understanding-mongo-query-behavior/
[12:48:07] <backSlasher> cheeser, any thoughts about my problem? Any way to trace mongo's writes to disk?
[12:48:27] <dragoonis> backSlasher, ok got it, i added an index on results_date field and then queried on it.
[12:48:50] <backSlasher> dragoonis, this means that it used the index. It should also scan less objects
[12:49:20] <dragoonis> The docs on indexOnly said ... "indexOnly is a boolean value that returns true when the query is covered by the index indicated in the cursor field."
[12:49:36] <dragoonis> and the only condition I have is results_date, so it should fully cover this index?
[12:49:42] <dragoonis> > db.company_category_score_months.find( { results_date: { $gte: new Date(1404169200000), $lte: new Date(1406761200000) } }).explain()
[12:50:45] <backSlasher> dragoonis, dunno but it tells mongo more specifically what you're trying to achieve
[12:50:58] <backSlasher> dragoonis, try using a direct comparison (A==B) and then see if indexonly lights up
[12:52:22] <backSlasher> I think indexonly is when mongo doesn't have to do any sacnning, and in your case you'll always have to scan something because you have no specific number, only a range
[12:52:47] <dragoonis> > db.company_category_score_months.find( { results_date: new ISODate("2014-07-01T00:00:00Z") }).explain()
[12:57:22] <dragoonis> > db.company_category_score_months.find( { _id: "53c3b5cb34d6199c7e8b465b" }, {_id:1} ).explain()
[12:58:40] <backSlasher> I think "only when the fields requested are in indexes used to find the records"
[12:59:01] <dragoonis> yep, i understand it now. So I need all fields back of course so indexOnly will always be false for me
[12:59:41] <dragoonis> Is having a BTreeCursor good enough for indexing (quickly scanning documents)
[13:00:23] <backSlasher> the real thing to note is to compare "n" (number of objects returned) to "nscanned" (or something similar), which is how many objects mongo manually compared
[13:00:42] <backSlasher> you have to keep the ratio high, meaning that mongo doens't scan objects you don't want
[13:05:50] <backSlasher> dragoonis, sure thing. I actually came here seeking help for my own problem... :-)
[13:17:33] <backSlasher> we are expiriencing really slow upserts (>3s) with a collection that holds medium (<10MB) files. They have "idhack", meaning the native "_id" index is used.
[13:17:33] <backSlasher> I don't have a very high lock percentage (~10%-20%) but I do see a very high write queue on my disk, once every ~20s.
[13:17:33] <backSlasher> AFAIK it's no the data flush because the console says that they are one minute apart. "accessNotInMemory" is lower than 1 for the relevant collection.
[13:21:00] <alanhoff> Hey guys, I'm willing to receive a JSON and use it as the a query param inside find(), is this ideia ok?
[13:46:11] <Cactusarus> can anyone help me figure out if I can use browseify with mongodb in node.js?
[14:28:30] <uehtesham90> hey, i want to make a copy of a collection in a database....there are two options: using copyTo() and cloneCollection....im not sure which is a better option
[14:34:09] <obiwahn> http://paste.debian.net/109545/ -- Is there a faster/better way to delete the homework with the lowest score?
[14:38:49] <obiwahn> i just think it is best to do stuff you can do in the db inplace instead sending stuff over the driver
[14:39:05] <pseudo_> I have a 3 node cluster that "ran out" of disk space. I am getting "Can't take a write lock while out of disk space" errors. However, each machine in the cluster has at least 900MB of disk space left. Why am I getting these messages still and how can I force mongo to delete some data?
[14:41:16] <rspijker> pseudo_: to get mongod to return disk space to the filesystem, either remove the files and resync from another node in the replica set, or run repairDatabase
[14:42:24] <pseudo_> rspijker: so if i remove the files directly from the filesystem, i can run repairDatabase and all the other data should be fine?
[14:43:42] <rspijker> pseudo_: no… you EITHER remove the files and sync back from another replica set member, OR you run repairDatabase
[14:44:52] <rspijker> however, if your DB files are already fairly unfragmented, which can happen if you don;t have many removals or updates. Then repairDatabase won’t give you much space back.. but then again, neither will anything else. Because you just have that amount of data
[14:45:41] <pseudo_> yeah, repair database didn't give a whole lot. i dont think i had a single removal/update
[14:45:51] <obiwahn> uehtesham90: lowest_score = min(entry['scores'][2],entry['scores'][3]) - ok works but not really elegant you could have matched for homework
[14:46:34] <pseudo_> each node in the cluster is short on disk space and the only place with 2gb of data is /var/lib/mongodb. is it safe to delete files straight from here?
[14:48:42] <pseudo_> rspijker: i have a 98GB database that I want to completely drop. I just don't have enough disk space to perform the drop operation
[14:49:55] <pseudo_> also, should i stop the whole cluster and remove it all at once? will the cluster be able to come back online if they all go down?
[14:51:51] <rspijker> if something does go wrong, at least you’ll have an intact copy of your data elsewhere
[14:53:14] <pseudo_> it looks like when I try to bring the node back online, it tries syncing the data but can't
[15:01:56] <pseudo_> rspijker: i just restored the files that i deleted. i took away the 0th file from the packets database. i am going to try removing a different one.
[15:02:45] <pseudo_> rspijker: i do, but i was just trying to free enough disk space to perform the drop command in mongo
[15:04:09] <pseudo_> so if i delete ALL the packets.* files, including the ns file - perhaps it will come back online wihtout the database?
[15:07:26] <dragoonis> Missing documentation on what 'multi' implies here: http://docs.php.net/manual/en/mongowritebatch.add.php
[15:08:48] <dragoonis> Derick, I believe upsert and multi need a 1 line description of each. I'm happy to add this now if you can approve my wording of 'upsert' and you tell me what 'multi' means :)
[15:09:40] <og01> If i want a replicaset with only 2 instances, how can I ensure that both instances become primary when network connectivity goes down?
[15:11:02] <og01> basicly I want both sites to continue to function even after network connectivity fails between the two
[15:12:31] <og01> we have two sites, one mongo instance in each, and a service running at each site collecting data,
[15:13:00] <og01> if network connectivity fails between the two, we want both sites to continue to accept writes
[15:14:44] <rspijker> there’s no way to do it through the regular replication/failover mechanism that I’m aware of
[15:15:08] <rspijker> maybe if you did some scripting around detecting it and then spinning up an arbiter or something
[15:15:14] <dragoonis> Derick, if I understand this right, 'u' is the updated data and 'q' is the query used to pull this record out ?
[15:23:10] <dragoonis> Derick, how does this look? https://gist.github.com/dragoonis/cc39aeb7b5cde622ced9
[15:24:28] <ejb> Is there a way to specify that id's should be strings not ObjectIds? Meteor uses strings so it doesn't play nice with documents that I've loaded via mongoimport
[16:34:31] <jParkton> Im having a couple of simple problems connecting to my mongodb via node I was wondering if someone could help point me in the right direction?
[16:36:49] <jParkton> Im trying to find my collections and return them I looked here http://runnable.com/UW3vaGTkq498AACC/query-for-records-in-mongodb-using-mongodb-native-for-node-js but its not working in my case maybe I am missing some package?
[17:05:49] <og01> im not 100% familiar with nodejs' driver, but it looks like to be that your connection isnt ready before you try to use it
[17:06:00] <jParkton> og01: Im pretty new to this whole thing, I guess the issue is in my getCollection there is no find method?
[17:08:12] <og01> jParkton: yest perhaps look at this tutorial: http://mongodb.github.io/node-mongodb-native/api-articles/nodekoarticle1.html
[17:10:58] <og01> infact its difficault to return anything meaningful at that point unless you use something like a promise
[17:11:41] <og01> right i can give you a short lession on event loops an async/non-blocking operations
[17:13:31] <og01> nodejs (and/or javascript in general) runs an event loop. this loop is native to nodejs (in some languages it must be implemented). This loops allows the programmer to setup watchers, listeners and fire events
[17:13:59] <og01> operations you perform in the event loop should be non-blocking, that is, code execution should never pause and wait
[17:14:38] <og01> instead it should setup callbacks/events that occur some time in the future (my description here isnt excellent, but hopefully you follow)
[17:15:24] <jParkton> I need to setup something in there to just find my collection and return it and stop executing
[17:16:18] <og01> so methods on MongoClient (and other places), take a callback (function) that gets called some time in the future with the results
[17:18:11] <og01> so with the connect command you only have access to the db object once the callback is called... you cant reliably use this object, and you can't just set it to a global variable because you wont actually know when its ready
[17:18:12] <jParkton> so I need to somehow setup a backend so I am not hard coding it all in my exposed files
[17:18:52] <og01> sorry if i go slow, it can be difficault to describe, and i want to make sure you get the whole picture
[17:21:00] <og01> now, with this understanding, you can work around things, you can set flags on variables when things are ready, setup timers and work wround things...
[17:24:00] <og01> MongoClient.connect() completes immediatly - it does not block, but the function you provide to it is called some time in the fucture when the function is complete
[17:24:26] <og01> but by that time you've already left your function (connectDatabase or getCollection)
[17:25:09] <og01> now I like promises, but some people I work with do not, i personally think some people simply fail to understand them.
[17:26:20] <og01> it is an object you can store, and you can call promise.then(function(db) {...do something with ...});
[17:29:15] <og01> it makes it so you can rebuild a more familiar flow to your code, you can return a promise of your results, be it successfull or failure, it makes returning a value usefull again, and also provides the equvilient of a thrown error via rejection possible
[17:30:11] <og01> I think you should perhaps grasp how to pass around callbacks through your code first
[17:30:40] <og01> and then later perhaps read up about promises, to see it that makes your life easier in the future.
[17:37:06] <og01> jParkton: http://0bin.net/paste/Eamh-ruAuGKYtSwX#xajgiczFiyAvbBR9-FZgiXeNEvBtyY3koT9f1pQKf2y
[17:38:12] <og01> jParkton: not tested, but i was trying to show how you can pass a callback to get the results from non-blocking async functions
[17:43:20] <og01> jParkton: http://0bin.net/paste/UnPq653TpdUyGl8Y#bkokqhmbfzunt+6KRCGIqNvOVhbGbRFpmUxV1UsPlaB
[17:43:59] <og01> please note that i wouldnt write code like this, but it is an example for you to learn from
[17:45:10] <og01> if it isnt already created, it connects stores the db object and calls the callback
[17:45:33] <og01> if it already is created then it calls the callback with the already existing db object
[17:46:41] <og01> as for your querying from your login page, you'd need to know your webserver/webframework and have an understanding of HTTP and potentially XHR/AJAX
[18:15:45] <apetresc> Hey, guys; I'm having trouble getting scons to respect the targets I tell it to build. Basically, even if I say something like "scons tools install --prefix=/whatever" it will insist on building and installing *everything*
[18:16:13] <apetresc> How do I convince scons to only build a subset of the targets? I only need to build the tools and/or client, not the server
[18:17:47] <apetresc> Or at least only *install* the client, even if it does get built as an intermediate step?
[18:38:26] <rickibalboa> Hi, I have a strange situation where a mongo query performed from a nodejs process hangs and ends up timing out. The database connection is made, as soon as I run any find() queries they hang, but they work if I use mongodb shell / robomongo.
[18:38:37] <context> with the aggragation framework can i $project an embedded documents field (this is an array of documents not a single embedded document)
[18:40:03] <apetresc> What do you mean, context? Like, you want to only project some of the documents in the array, or the entire field (which happens to be an array)
[18:41:07] <context> primary { blah []secondary }, many secondary, i want to project fields out of every embedded doc, some records only have one in the array, some have more
[18:41:53] <ranman> apetresc: scons mongo -j 8 or something like that seems to just build the shell for me
[18:43:28] <apetresc> ranman, not me :( I create a brand new directory (/Users/apetresc/tmp/mongo) and run exactly: `scons mongo install --prefix=/Users/apetresc/tmp/mongo`
[18:44:24] <apetresc> And then `ls ~/tmp/mongo/bin` returns everything... mongo, mongod, mongos, etc
[18:45:12] <apetresc> I feel like maybe even if the `mongo` target only builds the shell, the `install` target accidentally depends on `all`, perhaps?
[18:47:15] <ranman> apetresc: seems to be the case, I don't really want to go through the sconstruct atm though :/, my recommendation scons mongo -j8 then copy the build artifacts to wherever you want them to be
[18:47:54] <apetresc> ranman, sigh, I see :( If I track this down and submit a patch for it, there's no reason it wouldn't get pulled, right?
[18:48:51] <apetresc> So... what's the point of having individual fine-grained targets if it's by-design impossible to install any of them?
[18:49:45] <ranman> I don't think install was intended to be used that way -- nor do I think mongodb is meant to be installed on machines that way -- better to use the packages
[18:50:04] <ranman> if you're just doing development then you rarely need to "install" anything you just need to run tests
[18:50:47] <apetresc> No, but what if you're building it from source on a machine that just needs a client? You can't envision any scenarios where someone would want to talk to a mongo server without BEING a mongo server?
[18:51:17] <apetresc> Well anyway, I think I'll just workaround it by manually cp-ing the binaries out
[18:51:47] <ranman> apetresc: there are packages for the just the client, why are you building from source ?
[18:51:49] <apetresc> But for the record, I don't understand why installing a target doesn't install just that target :P But to be fair, I don't know too much about scons
[18:52:34] <apetresc> ranman, in this particular case it's because I'm trying to modify the Homebrew recipe for Mongo to support a --no-server option
[18:52:43] <ranman> apetresc: it's that scons and master and all of these things are for development of mongodb itself -- if you're using them your results are not guaranteed -- when most people deploy a database the want a battle tested release and package
[18:53:27] <apetresc> If i want to run the mongo shell on my mac, I have to install the mongo server
[19:00:39] <ranman> apetresc: if you're interested in trying to fix it I think the bug is in here somewhere: https://github.com/mongodb/mongo/blob/master/src/mongo/SConscript
[19:02:08] <ranman> apetresc: I don't have a lot of time to look into it right now but you have a good point
[19:22:34] <Gargoyle> Are there any channels for the university courses - can't see anything obvious. If there are no other suggestions, I'll be hanging out in #mongodb-m102 for the next seven weeks.
[19:33:45] <coreyfinley> My document has a field that is an array of dates where the given object meets some boolean requirement. so given a date range, could anyone help me craft a query that can find objects that have at least one date in the array that is within the given date range?
[20:17:16] <lfamorim> Hello! Why Mongo locked my database on indexRebuilder operation? I'm crying ... ;(
[20:19:39] <kali> lfamorim: well, for this kind of operation, you usually perform them by taking replica out of the replica set one at a time...
[20:24:55] <patricklewis_> meaning if I am viewing information on a document and I want to go to the next document how can I do so?
[20:25:26] <patricklewis_> a great example would be viewing a portfolio of web work. Each project is a document
[20:26:14] <patricklewis_> so if I am viewing a project and I want to view the next one based on when it was added
[20:27:24] <kali> patricklewis_: you need to have something to materialize this "added" operation. if you're using standard ObjectIds as _id, you're lucky, because they have a built-in timestamp you can use
[20:28:02] <patricklewis_> btw i just started using mongodb so Im using the wrong terminology let me know
[20:31:11] <patricklewis_> Basicly I am building a site/webapp using meteor js. I and the trailing part of the url is a parameter telling the app what document to pull the information from
[20:33:43] <kali> just look for the smaller value of the timestamp strictly higher than the one you're on
[20:50:03] <cozby> question... if an application is supplied one mongo url replica endpoint is that application aware of the other replica sets?
[20:50:20] <cozby> or does one need to supply ALL replica set endpoints for the application to be aware of them
[20:57:08] <kali> cozby: one is enough when everything is right. the driver will contact the one it know about and get the full list from there
[20:57:56] <kali> cozby: but if the one you're passing it does not work when the client starts, then you're in trouble. so it's better to provide the driver every replica you know about
[21:13:26] <sflint_> question for anyone : how does mongo choose which secondary to read from when using secondaryReadPreference?
[21:30:23] <lfamorim> kali, there was a possibility when I put rs.add({my-master-instance}), this replica remove all master information?
[21:51:33] <maxmanders> I know every mongodb configuration is different, but can anyone offer advice on useful values from mongostats to use as alerting thresholds?
[21:51:49] <maxmanders> I.e. if queues reads/writes are > X then potential problem; if > Y then significant problem etc?
[22:04:51] <maxmanders> joannac: I think unfortunately my answer will be 'give it a shot and find out' - I'm looking to put forward some useful indicative suggestions as part of a management offering. Would immediately go for MMS, but in this case it's not an option.
[22:05:29] <maxmanders> I just hoped there might be some indicative useful figures / ratios to look out for - but I really don't have all the details that I'd like to offer for a decent answer (own worst enemy)
[22:05:49] <maxmanders> joannac: among other things - not a reservation on my part at all, but client governance requirements
[22:08:06] <maxmanders> So as a starter-for-ten, nominally these values should be zero, and anything non-zero is worth investigating, and further tweaking as we understand more about the application and its behaviour?
[22:08:14] <joannac> but of course, you don't provision for your absolute peak load, so you will occasionally get spikes of queued readers and writers
[22:08:57] <maxmanders> so as reads or writes increase, would increased queue length necessarily inform a decision to add more replica sets to spread the load?
[22:11:09] <maxmanders> Er yeah :-\ iiuc scaling sharded replica sets can offer better performance, but premature sharding can be counter productive over having entire data sets on multiple replicas?
[22:14:01] <maxmanders> So unless there's a need to share to spread writes across different replica set members, would I be right in saying it would be good to start with having the entire data set on multiple replica set members? *with your assumption above*?
[22:18:21] <joannac> you can have: replica set A with all your user data, replica set B with all your invoicing data
[22:19:07] <joannac> or you can shard and have replica set A with half your invoicing and user data, and replica set B with the other half
[22:21:42] <maxmanders> And thought that one or more schema were present on a primary member, and were made highly available through other secondary replica set members.
[22:22:56] <maxmanders> Anyways - you've cleared things right up - and I've been awake too long - so thank you again, very much appreciated.
[22:40:34] <wc-> im running this on ubuntu, installed mongo via apt, it seems like the init.d script isnt stopping mongo
[22:40:47] <wc-> it will start it, but i have to send a kills ignal to the process id to get it to stop
[22:43:37] <wc-> ah i think i know, i might have the pid specified in a non-standard location in the mongod.conf
[22:56:29] <rasputnik> is there a downside to sticking each collection in its own db? trying to work around the db level locking
[22:57:33] <valonia> When i update an object, it changes _id and moves to the bottom, is there a way to fix that?
[23:00:00] <joannac> rasputnik: in my experience it won't help unless that's actually the bottleneck
[23:01:00] <valonia> $db->update(array('Code' => $_SESSION['user']['Code'], 'Username' => array('$regex' => new MongoRegex("/^$names\b/i"))), array('$set' => array('Position' => $i)));
[23:06:34] <rasputnik> joannac: everything is disk bound - but a db level lock doesn't sound like a good thing, we're pretty write heavy
[23:07:26] <rasputnik> joannac: yeah it's on the timetable for tomorrow. just wondering what the tradeoffs are
[23:08:12] <ashutosh> @jonnac when trying to delete any collection it throws "exception: too many namespaces/collections"
[23:12:22] <ashutosh> @anyone when trying to delete any collection it throws "exception: too many namespaces/collections"
[23:24:03] <temple3188> Hi guys I was hoping someone could help me with a complex (to my knowledge) query with mongodb
[23:28:03] <temple3188> I have the following collections in my db: https://gist.github.com/christemple/5cd17d9277308e760ad1
[23:33:52] <temple3188> I have updated the gist with what I am trying to achieve and can not figure it out :(