Log file Viewer
#mongodb logs for Monday the 1st of February, 2016
[05:13:12] <dddh> for example I have "user: 1, departments: {1, 2}" and "user: 2, departments: {2, 3}"
[05:13:56] <dddh> how to make them "user: 1, departments: 1", "user: 1, departments: 2", "user:2, departments: 2" and "user: 2, departments: 3"
[10:02:46] <fish_> should the localhost exception already get disable after I create a shard local user?
[10:05:00] <fish_> because I created a cluster via some scripts and before adding the first user worked just fine but now I get 'not authorized'... and IIRC the only thing I changed was adding creation of the shard local user to the scripts, so that happened before I attempted to add the cluster wide user
[10:22:12] <Lope> I really miss the writeresult messages that I get from the 3.2 mongo when using the 3.0 mongo. Debian Jesse only supports mongo 3.0 while Ubuntu supports 3.2.
[10:23:24] <Lope> As a fascinating turn of events, I was intending to set up an ubuntu VM, install mongo 3.2 on it and transfer the DB's across. But I accidentally added the ubuntu 3.2 to my debian jesse machine and then "installed" it. The amazing thing is it seems to work perfectly.
[11:08:39] <fish_> again reproduce the issue.. ater connecting to each of my shards + config server and creating a shard local user in the admin db with 'role': "root", I can't create a cluster wide user anymore..
[11:10:29] <fish_> is this expected? the documentation is a bit unclear on that. it says that 'The localhost exception allows you to enable access control and then create the first user in the system.' and 'To create users for a sharded cluster, connect to the mongos instance and add the users. Clients then authenticate these users through the mongos instances.'
[11:11:33] <Lope> m3t4lukas: I don't want bleeding edge. I just want write results in the terminal so I can see if my command went into a black hole, made a galaxy explode, or did nothing.
[11:11:40] <fish_> but it doesn't say if adding a local / shard / mongod user disables localhost exception for the cluster-wide / mongos user
[11:12:37] <fish_> or wait, maybe users in the admin db on the config server RS == cluster wide users?
[11:58:47] <m3t4lukas> fish_: https://docs.mongodb.org/manual/core/security-users/#sharded-cluster-users
[12:02:32] <fish_> m3t4lukas: but yes, it seems like it's how I described it. it also says 'MongoDB stores these sharded cluster user data in the admin database of the config servers.' - but it wasn't immediately clear to me that local shard users in the admin db on the config servers ARE the cluster wide users
[12:03:21] <fish_> but now I need to fix my actual issue.. the user creation stuff was just 'surprising' and I thought it could have been related
[12:03:43] <fish_> https://jira.mongodb.org/browse/SERVER-22392 <- this one, right now trying if I can reproduce it with 3.2.0
[13:44:16] <Lope> okay it was just a configuration issue. So I've got ubuntu mongodb repo working on Debian 8 Jessie. Should I be worried about future stability and updates? Should I move over to ubuntu?
[14:05:47] <fish_> can anyone tell me what exactly this error means? caught exception while doing balance: could not get updated shard list from config server due to ExceededTimeLimit Operation timed out
[14:29:04] <torak> Zelest: i am trying to use mongolab with my android app. Do i need a .jar file as lib?
[14:32:35] <torak> StephenLynx: i was using parse. and its going down so i have to move mongodb or somewhere else
[14:32:46] <StephenLynx> write a server application and have the client connect to the server and then the server work with the database.
[14:33:45] <cheeser> you don't want database credentials on the client nor connections open to the world.
[14:38:39] <StephenLynx> any other linux distro would work, but centOS is aimed at being a good server distro and you will have minimal versions on real servers.
[14:39:44] <StephenLynx> you can get one with 512mb RAM on digital ocean, but they are kind of crap, because you don't get swap by default.
[14:41:00] <StephenLynx> digital ocean is also pissy with some stuff. I asked them to open port 25 for me and they asked for a shitload of personal info, so I just moved back to linode after that
[14:50:28] <torak> StephenLynx: so still don't understand what should i do... Where can i find a documentation about this
[14:52:07] <StephenLynx> these are the most common and accessible tools, maybe not go being as accessible
[14:52:54] <torak> StephenLynx: yes. ı was set up an apache server. in settings you can select the port to listen
[14:54:01] <StephenLynx> on your server you handle these requests and work with the database based on what information you get from the request
[14:55:39] <torak> StephenLynx: i am a cs student and all the lab lessons we directly connect the project to the server with server id and pwd. But you say, in realworld its too nonsense. You need a 3. computer to handle this connections and it will be the only computer to connect the server. The other connect to this. For example everycodes work on the cloud and users app doesnt have any info about the codes work at server.
[14:56:45] <StephenLynx> I got mongo installed taking only local connections but without authentication
[15:12:45] <IanWizard> Is there a way to restore a all the documents in a collection (or one by one) to their previous revision? I may have just severely screwed up a collection and if I can have a script roll them all back, that would be much easier than restoring from a backup. All of the documents are still there, they've just had some junk values added to a few fields.
[15:13:27] <IanWizard> Everything I'm finding on google is for mongorestore, which seems to be for an actual resore rather than reverting a document.
[16:20:57] <chasepeeler> I'm having issues whenever one of my mongo nodes doesn't shutdown cleanly (which happened to all three nodes last week due to an issues with our VM hosts). When the server comes back up, mongod won't start. I've found the only way to get it to start is rm -rf /var/lib/mongodb/*, then start the services. At that point, I have to do rs.initialize() on the
[16:20:58] <chasepeeler> first node, and then do the same thing on the other nodes, returning to the first node and running rs.add() for each of them.
[16:22:45] <chasepeeler> is there any way to ensure that mongodb can start back up, even after an unclean shutdown, or, at least not have to go through the initialize/add steps every time it doesn't? We use mongodb to store session data for our application, so if its not working, neither is the application
[16:56:28] <torak> StephenLynx: so i have to create update drop all the tables with running queries??
[16:59:51] <StephenLynx> I know writing commands to handle with SQL dbs is laborious to say the least and trust me, there isn't that issue with mongo.
[17:00:05] <torak> but i need it while developing the app. So i can check if the object created deleted or etc. for trying things
[17:00:49] <StephenLynx> there is a pretty() function that you put after most commands that return data
[17:01:49] <StephenLynx> that has the same effect as JSON.stringify(object,null,2) instead of just running plain stringify
[17:03:43] <torak> StephenLynx: so. about the connection. i will connect to the server via the public ip in the app. So in the ubuntu i run some node.js code for doing do jobs and saving the data to server. Right?
[17:06:09] <StephenLynx> if you do that, you won't be able to use standard HTML forms with your back-end.
[17:08:41] <StephenLynx> because you can interpret the request as a single json object and then have sub-fields for stuff like authentication that is unnafected by parameters, because it is on a different base field
[17:11:34] <StephenLynx> not to mention when they start coupling stuff they shouldn't, like front-end and database.
[17:11:36] <torak> StephenLynx: so this is my graduation project. I have done good so far with parse.com, i still have time to finish it with parse because its going down at 2017. I can develop the parse version and graduate with it. But I dont want to create a crappy product for just graduating. But i am not sure if i go deeper and deeper. Or just use the firebase(something similar to parse)
[17:13:24] <torak> StephenLynx: yea but it has its own .jar lib for android and you can just use singinInbackground(username,pass) etc.
[17:14:16] <StephenLynx> so you can coup some benefit, since you are already too deep in debt to be worth quitting
[17:15:37] <StephenLynx> that sounds just like some crappy sandbox with extremely rigid framework stuff.
[17:15:50] <StephenLynx> if they have a FE lib for login, you are limited to what you have on the BE.
[17:16:04] <torak> StephenLynx: since i am not the expert on android-nodejs or mongo it will take too much time to finish it if i go with the mongo. I can finish it with parse but it will stop working in 1 year. so that makes me feel like i worked for nothing
[17:17:37] <torak> StephenLynx: thats why i need to move to somewhere else but i cant find a safe place
[17:20:07] <chasepeeler> whether college is necessary or not is one thing... but you can get a degree without being crushed with debt
[17:20:15] <torak> StephenLynx: i probably will have 10k dollar debt for me and my girlfriend combined
[17:21:14] <torak> StephenLynx: Goverment gives you credit for living etc while youre a collage studen (if you want) and you pay back when you find a job
[17:21:53] <StephenLynx> here in br you can get a funding for your college on private institutions, but its actually free.
[17:23:05] <torak> StephenLynx: yeah we have that here too but there is not too much people to support students. they generally prefer drinking more coffe and eating more food than sharing
[17:44:24] <deathanchor> working in college helped me prep for working in the real world. which turns out that 90% of time everything is broken :)
[18:51:44] <supersym> anyone here with sufficient knowledge of the (node/js) driver to answer the question why/what the reason for ObjectID and ObjectId is...
[18:53:06] <nikitosiusis> how can I find out that my mongodb stopped fitting into memory and started doing some reads from disk? is there some metric in mongo statistics for this?
[18:53:26] <supersym> is cheeser a bot or what? always helpful answers here... seems to me, this might be the first logical place really, and not so much node
[18:53:51] <supersym> somehow I get the impression the first answer always is "ask someone else" .. no disrespect
[18:55:27] <supersym> well there is a ObjectID and ObjectId currently in the driver, I've always assumed that the former was a fek-up by some sleepy head, and left in for backwards compatible reasons
[19:01:19] <supersym> hmmm... ok it seems recent version, only has ObjectID left... ugh. Now I'm really confused to see who's bringing in the ObjectId.. but I guess it answers which one is the 'right' one
[20:06:25] <zylo4747> i'm trying to install a new 3 node replica set on linux and it's just hung and the error logs are filled with "not authorized" message. can someone help?
[20:27:21] <m3t4lukas_> zylo4747: "not authorized" can have many meanings and without any logs one can only guess. My guess is you are having security enabled but you did not provide any keyfiles. https://docs.mongodb.org/manual/core/security-internal-authentication/#inter-process-auth
[20:57:26] <chasepeeler> given: mongodb://mongo1:27017,mongo2:27017,mongo3:27017/?replicaSet=espsession&journal=false&w=0 for connecting to mongodb, via PHP, what happens if, after a crash, the three servers come up with the replicaset not initialized? will it still connect to the first one listed and read/write to it without issue? Or do I need to connect to a single server without
[21:12:48] <gruntruk> hi, i was wondering if someone could help shed some light on understanding `.explain()` and indexes… am helping out do some performance debugging on a legacy app, it has a pretty large collection with a few queries that are performing rather atrociously. after adding an index to match query pattern (filtering on three+ fields), i still only see it hitting the index for a single field. any pointers? this query in particular is quite slug
[21:14:11] <gruntruk> so, given i’m querying on fields: “company”, “user_id”, and “status_code” - even after adding an index for company, user_id, status_code, the explain only shows it using “BtreeCursor user_id_1”
[21:33:43] <Freman> more of a middle end - we have a web ui that people can query but they keep bypassing it and connecting directly - and if we just close the port then someone will cry to management and management will get crabby to us
[21:35:01] <Freman> it makes direct queries that support indexes, but people connect directly and robo mongo is constantly leaving their dumb queries running when they quit
[21:35:26] <Freman> but writing something that takes the given query, checks to see if it uses any indexs before permitting it to run. much sexy
[21:36:45] <Freman> Personally, I'd delete over half the data, drop a firewall over the lot of it and tell them to use the application or bugger off
[21:37:19] <Freman> but I don't get to make that decision, I just have to deal with the consequences of people running dumb things.
[21:37:56] <Freman> "why is our logging router backlogged by 1000000 entries?" cos some fool ran an unindexed query on storageSize" : 232634284448 - log_data
[21:38:29] <Freman> I'll write the middle-end, it'll let them run their queries only after they include indexes in them
[21:39:39] <Freman> why we keep all this crap... or log it in the first place "remote api spat out a 500 error so instead of logging the error message and the key components that let us replicate the request, we're going to log the 8k of data that makes up the entire request and response"
[21:40:44] <Freman> so while I might sound frustrated with mongo, it's only with mongos/balancer. it wouldn't even be a problem if we weren't being dumb in the first place
[21:41:36] <StephenLynx> blame the dev that didn't just said "that is a greaaaat idea, but sorry, nope, that is literally impossible to be done."
[21:42:10] <Freman> any time there's a backlog on the logger I just log in and kill any queries that have been running for more than 10s
[21:44:20] <Freman> iptables -I INPUT -p tcp --dport 27017 ! -s 10.10.10.10. -j REDIRECT --to $middleend
[22:15:17] <gruntruk> Freman: wish the developers had “Query Guard” prior… this is the first project i’ve used mongo on… it’s quickly becoming apparent that the resident “experts” have little clue as to how they have been using it all this time.
[23:04:52] <uuanton> does anyone know how to connect replica set without resync data. Two instances made of same data snapshot but somehow drop everything when sync