Log file Viewer
#mongodb logs for Tuesday the 14th of April, 2015
[00:13:20] <GothAlice> FrenchieUser12: Note that "collection" in my examples is meant to be the name of whatever your actual collection is; the one storing the data you are trying to retrieve. If you are find_one'ing on a collection different from the one you are inserting into, you can't expect it to work. ;)
[00:13:45] <GothAlice> (In the case of my interactive shell example, the name of the collection I inserted into _and_ read from was "collection". Apologies if that was unduly confusing.)
[08:44:56] <jeanre> why would total be 0 in this case http://pastie.org/private/xuunqlqmjns2nhh0jfj8a
[13:09:00] <brano543> hello, i am trying to update document and i want to add a field pois with array of objects but with no success,nothing gets added. if i try to add just array of integers,it works,but not with array of objects. I am doing BSON("$set" << BSON("pois" << pois.arr())) Can anyone help me?
[13:11:31] <brano543> StephenLynx: i am using c++ driver. full syntax mongo.update("data.vrcholy", BSON("_id" << n_id), BSON("$set" << BSON("pois" << pois.arr())) );
[13:12:15] <StephenLynx> have you tried to perform your operation on the shell first to make sure your logic checks?
[13:18:06] <brano543> StephenLynx: this works in console db.vrcholy.update( {"_id":3039828758},{$set:{"pois":[{"test":"testval"}]}})
[13:19:47] <brano543> StephenLynx: i don t know what am i doing wrong,i am creating array of objects,see https://dpaste.de/fFWH
[13:22:33] <brano543> StephenLynx: strange thing is when i created objects exactly the same and used isnert it got inserted properly.....
[13:54:00] <brano543> StephenLynx: its strange, i am creating exactly same object as in console and it doesn´t update it
[13:55:50] <StephenLynx> more than often lower level tools will require you to learn their intricacies
[14:53:42] <StephenLynx> brano543 I doubt it, that is a basic operation, to set a field with an fixed array.
[14:55:30] <brano543> StephenLynx: this is ridiculous,i have even tried to specify the multi option and it still doesnt work i have followed this http://www.slideshare.net/YnonPerek/mongodb-for-c-developers
[14:56:28] <brano543> StephenLynx: i have read everythin i have found on that side and on official mongo site
[14:56:47] <StephenLynx> there is a reason lower level languages are not on the bottom shelf of most programmers ¯\_(ツ)_/¯
[15:21:16] <adrian_lc> hi, how can I increment the document with the lowest value for a particular field, atomically if possible?
[15:21:39] <brano543> StephenLynx: it seems now its inserting somethign,but sometimes it inserts empty array,weird,i am sure every time is not empty
[15:34:29] <brano543> StephenLynx: you know , i need to optimize my importer to DB,because i can´t keep all the data in memory, so i insert into DB and then query
[15:53:59] <brano543> StephenLynx: can you please explain me the WriteConcern? Somehow i don´t see an option there to report me if any errors happened during the bulk update operations
[15:56:54] <StephenLynx> I use the node.js driver, it returns me an error object as the first parameter of the callback.
[15:57:35] <brano543> StephenLynx: i hate node.js, someone helped me to get it working here last time
[15:58:20] <brano543> those callbacks its a mess, i had to use some async library to get the stuff executioned when i needed
[15:59:15] <brano543> well,i couldn´t get it working without it,it closed the communicatino socket while i still needed to perform soem queries
[16:02:03] <brano543> my problem is i don´t understand how it works,i saw soem examples, someone even told me why do i call connect DB on every request,answer because i don´t understand this :( i dont know how to create global variable connection,there make connect to DB and for everz querz call it like i do it in c++ https://dpaste.de/q4hi
[16:03:03] <StephenLynx> you don't understanding something doesn't mean the technology is bad or messed. if you have discipline, callbacks are no more complicated than other forms of abstraction, such as interfaces or multiple inheritance.
[16:05:42] <brano543> can you show me some example please? my project is routing application, i have node.js server which processes queries and communicates with c++ wchich actually searches the graph and sends back JSON
[16:07:27] <StephenLynx> javascript has built-in json support, you literally has to do nothing to add a js object to mongo.
[16:08:12] <StephenLynx> with javascript you can just initialize an object using var myObject = {a:'banana', b: [1,5,6]}; and then add myObject to mongo.
[16:08:19] <jaitaiwan> Hey StephenLynx... awhile back you sent me a link to one of your projects which contained awesomely formatted JS and a suggestion of a book to read. Any chance you remember that stuff still?
[16:08:42] <brano543> StephenLynx: hmm,but even if the graph is stored in DB and zou can query nodes, you don´t have fast enough implementations of priority queue compared to c++
[16:10:03] <jaitaiwan> StephenLynx: I can't remember but it had something to do with not writing bad code because Bad code is what causes callback and promise hell
[16:10:12] <StephenLynx> and if you are using a graph, then perhaps mongo is not the best tool for the jobs, there are dbs dedicated to graphs.
[16:10:49] <StephenLynx> unless you are writing native code or using native modules, it is a drop in replacement.
[16:13:44] <brano543> StephenLynx: thank you, i will definitely check that out. i see, io.js works the same way as node.js, i updated the dependencies with npm update anyways
[16:15:13] <StephenLynx> for you I can direct you to https://iojs.org/en/index.html and https://github.com/iojs/io.js
[16:16:29] <brano543> StephenLynx: thanks,i will need a time to process it. for now, i am trying to make those updates i struggled with today as bulk operation :)
[16:19:43] <StephenLynx> running a windows server is a laughably bad idea, and you are better developing in the environment that will be the one on the server.
[16:20:00] <StephenLynx> so not much point into headbashing to make stuff work properly under windows.
[16:21:03] <brano543> StephenLynx: its fine for me, the importer will work on windows,its easier, the server may work on windows. I am not experienced enough to provide all the dashes for clients needed to compile the dependencies under linux :D i am just a student haha
[16:23:03] <StephenLynx> do you actually have a benchmark to prove that C++ will provide a gain in your case?
[16:24:49] <brano543> StephenLynx: i am also thinking about on-the-flz approch,that the function succesor(node) will query DB so i don´t need to keep the graph entirely in memory
[16:25:30] <StephenLynx> will your C++ code free the RAM while it awaits for the IO operations to finish?
[16:27:15] <StephenLynx> if you expect a load big enough for that to matter, then the RAM should matter too.
[16:28:15] <StephenLynx> have you tried to implement the A* search in js and benchmarked it against the C++ code?
[16:29:22] <brano543> StephenLynx: i will try it, first i need to focus on the matters which is my bachelor thesis about
[16:35:36] <StephenLynx> but again, as everything that is important, will not work very well on windows because it is an useless pile of radioactive trash.
[16:37:03] <StephenLynx> but you need to store data of cars and people, create a collection for cars and a collection for people.
[16:39:24] <PedroDiogo> for this, I'm using a single Schema on my nodejs app for messages only with from, to, message, timestamp and lastRead parameters
[16:40:53] <StephenLynx> do you have accounts for the users? or you just trust they are who they clam to be?
[16:41:08] <PedroDiogo> so: one "schema" is fine, but I should use mongodb's official nodejs driver + io.js - right ?
[16:44:34] <PedroDiogo> I just think socket.io is easier to use for this single purpose which is a simple chat application - thats all
[16:44:55] <StephenLynx> doesn't socket.io requires you to use a specific library on your front-end?
[16:49:46] <StephenLynx> and I have another project developed at my company that is exactly what you are doing, a 1x1 chat.
[16:49:57] <PedroDiogo> it is a real time web app with a REST web service bellow where I need to pass and receive special JSON to different services and the front end as well
[16:51:02] <PedroDiogo> yeah I know, but I actually like the express way of doing things with modules, routes and all that. It eases the workflow with my colleagues
[16:51:41] <PedroDiogo> i understand it may slow things down a bit (I really do - I'm new to node.js and mongo, but not a new programmer), but, for this project, I must do so
[17:01:47] <StephenLynx> and to update the readme, it still says node is a dependency instead of io.js
[17:05:57] <PedroDiogo> I also want to migrate from orchestrate.io to my own MMS solution... do you recommend any server to do so?
[17:07:58] <StephenLynx> it creates a clusterfuck so you have to type a little less and makes your code dependent on it forever so you can say you use hip technologies which is not a problem since you won't have to maintain the project after you lose your job.
[17:14:41] <StephenLynx> if you host it on a regular dedicated server, it is just a mongo server running.
[17:19:37] <StephenLynx> to put things in perspective, see how much it costs to store a single terabyte with linode.
[17:21:48] <StephenLynx> with linode you will be paying thousands of dollarts per terabyte per month
[17:39:13] <PedroDiogo> I can always transfer my data to a cheaper alternative when using MMS, right ?
[17:42:31] <PedroDiogo> this is mongoose's schema: https://gist.github.com/PedroDank/f2932c0f1f297da8437f
[17:58:48] <Diegao> hi, I would like to compare two collections (one original and its copy) to get the different records
[17:59:27] <Diegao> these two collections were generated in different scenarios, so I would like to see the difference between then
[18:10:32] <chairmanmow> I'm trying to update a record and push to two arrays inside the document. I've done it successfully with one array, but the second one doesn't get pushed, and mongo doesn't throw an error. Is there anything glaringly wrong with this syntax ? : Item.update({_id:updateQuery.item._id},{$push:{logs: updateQuery.logToPush, assignments:updateQuery.assignmentToPush}}
[18:44:00] <StephenLynx> chairmanmow it is failing to update a second document or the second array in the document?
[19:32:48] <GothAlice> I find it handy to force people to slow down and watch instead of throwing a wall of code at them in a gist.
[19:34:42] <GothAlice> Unfortunately to get there it literally has to replay every step to build up the final result; it's not like a video file with keyframes, here.
[19:37:29] <StephenLynx> if people just alt+f4 and start demonizing the site, then it will never take off.
[19:52:43] <GothAlice> (Saying something is unbearably slow when the whole point is for it to be slow a good argument does not make.)
[20:06:37] <pjammer> if mongod process was using 75%+ cpu and then dropped to under 1%, yet currentOp() didn't have anything in it... can i find out why the sudden drop occured?
[20:07:08] <GothAlice> pjammer: If you have "slowMs" profiling enabled, the query/op that caused the spike may appear in your logs.
[20:07:29] <GothAlice> If you don't have that profiling level (I believe it's 1) enabled, it likely won't show.
[20:08:42] <pjammer> it's been sucking that cpu % since the weekend, after we killed some Index job (long story) but i've been checking currentOp for days and it's been blank with no queries.
[20:09:16] <GothAlice> Aye; a background index operation would show up in the list of active operations… if that's what it was doing.
[22:04:53] <GothAlice> harttho: http://docs.mongodb.org/manual/reference/method/db.collection.stats/ will get you stats by collection, http://docs.mongodb.org/manual/reference/method/db.stats/ for a whole database at once.
[22:06:24] <harttho> GothAlice: Thanks, I guess I do see the reason that another level up doesn't exist haha
[22:06:24] <GothAlice> If you're looking for the on-disk size of all, you can just sum those up from the files on disk. (du --total) Otherwise, you'll have to iteratively collect each DB's stats and aggregate.
[22:51:12] <pjammer> Single master... when is the right time to switch to replication if you want reads to always be available when called from your web app. i have stats to aid the convo too
[22:57:27] <pjammer> if an existing /data directory is owned by root, will changing ownership to mongodb, mess it up, for lack of a better word?
[22:58:13] <GothAlice> You'll need to restart the process as the correct user anyway, so step 1: stop the mms processes, step 2: chown, step 3: re-start as the mongodb user.
[22:58:32] <GothAlice> Replace step 1 with: stop the existing process if MMS is not yet installed. ;)
[23:10:31] <pjammer> so in mms Btree and Journal Stats both took nosedives, and then my Queues, Opscounters,Background Flush Avg all went down to normal or 0 levels.