pmxbot IRC Log Viewer

[00:01:28] <dburles> thanks for your help @joannac

[00:03:46] <joannac> np

[00:43:59] <mekhami> can i have a model field field that is a computation based on other model fields?

[00:51:21] <joannac> mekhami: sounds like a mongoose question?

[00:52:43] <mekhami> joannac: yep but idk is there a mongoose question?

[00:53:49] <joannac> you might get a better response in the mongoose support channels http://mongoosejs.com/

[01:00:01] <duallain> is it possible to enable client authentication without internal authentication in a sharded replicaset?

[01:00:11] <joannac> no

[01:00:27] <joannac> how will the members talk to each other?

[01:01:14] <duallain> @joannac Thank you for the quick answer!

[09:02:10] <stangeland> hi. i am creating a datamatrix every second on a multiprocessor system. the datamatrix is essentially a float array which could be saved as a BLOB with an attached timestamp. My question is the following: If i have a lot of cores creating such datamatrices each second, will mongodb be able to efficiently store them and handle the synchronization?

[09:54:59] <oskie> zero downtime manual failover is not possible with MongoDB 3 replica sets, is it possible with sharding by adding a new shard to replace an old one?

[09:57:09] <Lujeni> Hello - there is a way to limit the memory utilization of mongodb ? thanks

[09:57:38] <oskie> Lujeni: you can always try ulimit

[09:58:14] <Lujeni> oskie, do u know what happened when the limit is reach?

[10:00:22] <oskie> I don't know how MongoDB would react, but probably the same way as if the memory of the machine was full

[10:01:46] <Derick> why do you want to limit the memory?

[10:02:01] <Derick> and I doubt ulimit is going to do what you want with the MMAP storage engine

[10:03:01] <Lujeni> Derick, i use the wiredTiger engine

[10:03:48] <Lujeni> mongodb take 90/95% of the memory. imho it's pretty risky

[10:04:06] <Derick> memory is good for a database, don't try to outsmart it

[10:05:49] <Lujeni> i know

[10:06:55] <Lujeni> Derick, however you don't think that limit the memory of a database could be usefull ?

[10:07:13] <Lujeni> i don't want to use a cgroup :_

[10:12:31] <Derick> Lujeni: don't OS limit it - configure the database to use less memory if you really *must*

[10:15:05] <Lujeni> Derick, ok what's the best way ? tuning the cachedSizeDB ?

[12:11:48] <_KryDos_> Hello. Could please help me with next question. Collection in my database has 6M of records. I'm trying to get some information from this collection for one month (it's about 3M of records). I'm using aggregation and execution time of the query is about 11 seconds. All fields involved in aggregation query have indexes. I just want to know if 11 seconds is bad result or not? and if you have any suggestions how I can impro

[12:11:48] <_KryDos_> ve it, that would be great.

[12:14:55] <_KryDos_> It's actually not so smart question, since I didn't explain what kind of hardware I have and what exactly my aggregation query do... but anyway... maybe you have some experience or examples that you can share. I will try to compare your examples with my result.

[14:45:29] <the_voice_> Pretty new to mongo here so sorry if this comes across as a stupid question

[14:45:46] <the_voice_> I am using Meteor so I am learning mongo too

[14:46:23] <the_voice_> If I am doing a comment and a reply system for a page. Would I be best to have one document for each page that contains all of the comments and then a subdocument with replies

[14:46:28] <the_voice_> or should it be one doc per comment

[14:48:23] <StephenLynx> what is meteor?

[14:50:03] <the_voice_> meteors

[14:50:08] <the_voice_> full stack javascript framework

[14:50:10] <the_voice_> that uses mongo

[14:50:38] <StephenLynx> sounds bad.

[14:50:54] <the_voice_> based on what?

[14:51:01] <StephenLynx> bloat

[14:51:09] <the_voice_> meh, it's irrelevant to the question

[14:51:28] <StephenLynx> not really, does it abstracts the database in any shape or form?

[14:52:20] <StephenLynx> I mean, more than the native driver does?

[14:54:06] <the_voice_> Meteor is really an abstraction of node.js

[14:55:42] <the_voice_> You write regular mongo commands on the server side to access the database. What meteor does is it makes everything realtime, so it will for instance read the oplog(if you have one) and automatically update the client side when you make changes. If you re not using oplog then it polls the database automatically(which is not ideal)

[14:56:43] <the_voice_> On the client side you have minimongo which is basically a JS version of mongodb that runs on the client. So you can also write most mongodb queries on the client side.

[14:57:09] <the_voice_> Security as to what the client can see is handled by a publish/subscribe system(you choose what to publish to what client)

[14:57:49] <the_voice_> Pros: Insanely rapid development, one language, less lines of code do more.

[14:57:52] <StephenLynx> omg

[14:57:53] <the_voice_> Cons: Nightmare to scale

[14:57:55] <StephenLynx> that is like

[14:57:59] <StephenLynx> the worst thing I ever read

[14:58:08] <StephenLynx> I would kill myself before touching that.

[14:58:15] <the_voice_> You are very much so entitled to your opinion on that

[14:58:34] <StephenLynx> I can`t even figure out what the hell is going on their repos

[14:58:53] <StephenLynx> https://github.com/meteor/meteor/tree/devel/packages wtf

[14:59:01] <StephenLynx> this is probably the worst framework I ever seen

[14:59:03] <the_voice_> Again, you are very much entitled to your opinion on that

[14:59:24] <bogn> as he said, it's an abstraction on node.js a whole application eco-system

[14:59:24] <the_voice_> It has lots of issues, but it does one thing very well. Lets you write programs insanely fast that would normally take months

[14:59:44] <bogn> , a whole application eco-system

[14:59:53] <StephenLynx> quality > quantity

[15:00:07] <StephenLynx> it doesn`t matter if you crap out awful software insanely fast.

[15:00:08] <bogn> I wouldn't agree it's insane and worst framework

[15:00:09] <StephenLynx> it still crap.

[15:00:18] <StephenLynx> dude, did you see that link?

[15:00:22] <StephenLynx> this thing is bonkers

[15:00:25] <the_voice_> That's great StephenLynx

[15:00:29] <StephenLynx> that`s awful

[15:00:39] <the_voice_> you can think it is bonkers, but here is the most important thing you need to understand

[15:00:45] <the_voice_> 99% of startups never scale, they die before scaling

[15:00:52] <StephenLynx> >muh scale

[15:01:00] <StephenLynx> >throwing buzzwords

[15:01:09] <StephenLynx> if you don`t want to have an argument, don`t.

[15:01:12] <bogn> they die before they have enough users

[15:01:15] <the_voice_> Bingo

[15:01:37] <the_voice_> so Meteor let's me put out an MVP in a month. I can still scale it to above a 100K users without a problem and at that point I can raise real money and go to a real framework

[15:01:52] <bogn> so, the thing with embedding or not is, will you need those embedded docs on their own often

[15:01:54] <StephenLynx> I wouldn`t be surprised how many of them die because they go with crappy software made by incompetent and lazy developers using crap frameworks.

[15:01:56] <the_voice_> Meteor is answering a very important need.

[15:02:11] <the_voice_> That is the dumbest comment i have ever read

[15:02:14] <StephenLynx> no, meteor is just "shitty framework n 23423"

[15:02:25] <StephenLynx> it is even worse than express

[15:02:26] <the_voice_> no user gives two shits about the software running their applications, they care it works

[15:02:34] <StephenLynx> at least express doesn`t bundle db and front-end on it.

[15:02:40] <the_voice_> if it works, they are happy. For all they care it's a bunch of monkeys pushing button

[15:02:50] <the_voice_> Case in point, twitter was built on Ruby On Rails which scaled like shit back then

[15:02:58] <StephenLynx> it won`t work for long when the framework starts crapping itself and putting a ceiling on the developer

[15:03:11] <StephenLynx> because web frameworks are all crap.

[15:03:15] <the_voice_> They rewrote their backend in Java and C and well they are doing fine

[15:03:19] <bogn> StephenLynx it solves real-time updating on all clients for people who need that. Think of a collaborative CMS or what not.

[15:03:21] <StephenLynx> ruby on rails, laravel, meteor

[15:03:39] <StephenLynx> that is like, 2 hours work.

[15:03:39] <winem_> Stephen, which stack do you prefer for web applications?

[15:03:45] <StephenLynx> node

[15:03:50] <the_voice_> Use node all you want

[15:04:06] <StephenLynx> I wast not answering you.

[15:04:15] <the_voice_> You'll take a year to release something while the guy writing in Meteor will have released three different versions getting feedback from real users

[15:04:15] <bogn> (15:56:51) bogn: so, the thing with embedding or not is, will you need those embedded docs on their own often

[15:04:35] <the_voice_> Well, they are replies, they will never be on their own

[15:04:39] <the_voice_> they will always be with the comment

[15:04:47] <bogn> most popular replies page

[15:05:00] <the_voice_> Nah, not going to happen

[15:05:02] <bogn> or widget on the front page

[15:05:10] <the_voice_> Nope, it's private to that page

[15:05:23] <bogn> then embed it

[15:05:43] <the_voice_> Does it matter that it makes the embedding two deep?

[15:05:46] <winem_> can I interrupt the discussion with a dumb question regarding handling users in replica sets? did not work with mongodb for 8 months and can't remember how I did it in prior projects

[15:06:04] <the_voice_> I mean you have the page which contains an array of comment documents, and then you have the replies inside the comment

[15:07:15] <bogn> the_voice_: https://docs.mongodb.org/ecosystem/use-cases/storing-comments/

[15:07:23] <bogn> they documented that quite well

[15:07:39] <bogn> https://docs.mongodb.org/ecosystem/use-cases/storing-comments/#embedding-all-comments seems to be what you're asking

[15:07:46] <the_voice_> Thanks

[15:07:53] <bogn> I also don't get the distinction between comment and reply

[15:07:57] <bogn> aren't they both comments

[15:08:41] <the_voice_> Yah but a reply is to a comment

[15:08:50] <the_voice_> so it is embedded in that comment

[15:09:09] <bogn> they used replies for both kinds

[15:09:17] <bogn> reply to article and to comment

[15:09:28] <bogn> or rather reply to reply

[15:09:32] <the_voice_> I will read it over

[15:09:34] <the_voice_> thank you

[15:09:56] <StephenLynx> power peak

[15:09:59] <bogn> ah forgot that one: If, in the future, you want to switch from chronological to threaded or from threaded to chronological, this design would make that migration quite expensive.

[15:10:27] <StephenLynx> <the_voice_> You'll take a year to release something while the guy writing in Meteor will have released three different versions getting feedback from real users

[15:10:28] <StephenLynx> opinions.

[15:10:54] <the_voice_> Okay StephenLynx... You are entitled to your viewpoint

[15:10:54] <StephenLynx> I have worked on complex systems before, I would take a week at best.

[15:11:02] <the_voice_> Good for you

[15:11:09] <StephenLynx> hurr durr ur entitled to ur opiniums derpderp

[15:11:17] <StephenLynx> dont try to take the high road if you dont have any arguments

[15:11:23] <StephenLynx> you already said that 3 times.

[15:11:30] <StephenLynx> ./ignore StephenLynx

[15:11:35] <StephenLynx> without the .

[15:11:39] <StephenLynx> your are welcome

[15:11:45] <Zelest> oh, drama :o

[15:11:48] <Zelest> *grabs popcorn*

[15:11:50] <the_voice_> you're

[15:11:51] <the_voice_> lol

[15:12:05] <StephenLynx> yeah, typo

[15:12:14] <the_voice_> No drama. I have too much work to do. StephenLynx is enjoying bashing a framework for the sake of it

[15:12:29] <the_voice_> They are all tools that give you answers to certain problems, some are good for something, some are good for others

[15:12:30] <StephenLynx> i am bashing a framework because its bad.

[15:12:34] <the_voice_> Okay

[15:12:39] <StephenLynx> some are plain bad.

[15:12:42] <StephenLynx> like PHP

[15:12:46] <StephenLynx> or VB

[15:12:55] <the_voice_> Facebook was built on PHP

[15:13:02] <StephenLynx> no, it was built on C++

[15:13:03] <Zelest> Facebook is bad.

[15:13:08] <the_voice_> seems to be doing pretty well when I last checked, so is wordpress which powers 80% of the web

[15:13:12] <StephenLynx> it only had a PHP interface.

[15:13:16] <StephenLynx> HAHAHAHAHA WORDPRESS

[15:13:22] <the_voice_> no StephenLynx, Facebook was originally built on PHP

[15:13:26] <StephenLynx> >80%

[15:13:27] <the_voice_> lateran it was moved to C and other stuff

[15:13:32] <StephenLynx> 25% at best.

[15:13:39] <Zelest> This convo is soooo related to MongoDB :D

[15:13:43] <StephenLynx> and anyone that worked with wordpress knows how much it sucks ass.

[15:13:50] <StephenLynx> hey, I wasn`t the one who started it.

[15:14:00] <the_voice_> You are correct 20% of websites are wordpress

[15:14:08] <StephenLynx> >hurr it just werks

[15:14:15] <the_voice_> yes you did, you asked me what I was coding in and I said Meteor, you then proceeded to bash it like crazy

[15:14:18] <StephenLynx> is just an excuse used by incompetent people

[15:14:25] <StephenLynx> to make crap without any regards to quality

[15:14:34] <the_voice_> personally I don't care what someone writes in as long as they get the results I need

[15:14:42] <the_voice_> for all I care write it all in basic

[15:14:42] <StephenLynx> you take that software that "just werks" and think about it again a year down the road.

[15:14:45] <the_voice_> it doesn't matter

[15:15:04] <the_voice_> StephenLynx, what costs more money usually, hardware or coders?

[15:15:39] <StephenLynx> developers. and you will have to waste a lot of man power fixing the software deficiencies when you start with a "just make it work" mindset.

[15:15:39] <the_voice_> I can scale the shit out of most of those shitty frameworks cheaply using AWS

[15:15:53] <StephenLynx> >scale

[15:15:58] <StephenLynx> >throwing buzzwords again

[15:15:58] <the_voice_> hell Instagram sat on 100s of instances and had 11 people running the whole thing. All written in Django

[15:16:18] <the_voice_> Yes, but I need to first know that my product is worth the manpower to waste

[15:16:40] <the_voice_> So I can spend a year building something or I can build an MVP quickly that works and see what the users think of it

[15:16:41] <StephenLynx> you don`t even try to make stuff without frameworks.

[15:16:47] <the_voice_> the only thing that ever matters is the User

[15:16:48] <the_voice_> nothing else

[15:16:52] <StephenLynx> how much experience you had with node before using frameworks?

[15:17:00] <the_voice_> Why does it matter?

[15:17:03] <StephenLynx> how much you know how the framework operates under the hood?

[15:17:08] <the_voice_> Why do I care?

[15:17:23] <StephenLynx> because that is the same as giving a child a chainsaw

[15:17:25] <the_voice_> Your mistake is you think the technology is important

[15:17:30] <the_voice_> the tech is not important

[15:17:34] <StephenLynx> :v

[15:17:42] <StephenLynx> well, if you are just a code monkey

[15:17:49] <StephenLynx> I don`t have anything else to say to you

[15:17:53] <the_voice_> It's a tool

[15:18:12] <the_voice_> that is all it is.. I make products for my users and it has to give my user the best experience possible

[15:18:41] <bogn> he's a startup guy, that's trying to build prototypes fast, and if the prototype works out, he'll rewrite it with the right tool and resources to afford more dev power

[15:18:47] <the_voice_> bingo

[15:18:50] <the_voice_> doesn't matter, he left

[15:19:02] <the_voice_> also my wife is Interactive designer(UI/UX etc..)

[15:19:16] <the_voice_> so I believe in the user...

[15:19:31] <the_voice_> Also Learned the hard way previously that working a year on something no one wants is a pretty big waste of time :)

[15:20:52] <the_voice_> Honestly I learned so much from my wife when I met her like three years ago. I had started to learn it previously from a failed venture, but then seeing a UI/UX persons perspective changed everything for me

[15:21:10] <bogn> regarding the MongoDB docs I can tell you, that they are quite extensive and provide useful tips for use cases and such (as you saw)

[15:21:33] <bogn> and there's also https://www.mongodb.com/presentations/

[15:21:46] <bogn> also a wealth of insights

[15:22:07] <the_voice_> Thanks, I was always a mysql guy. So I am now really trying to learn how to use mongodb correctly

[15:22:22] <bogn> above is a link to all presentations, that page has a search bar as well

[15:22:39] <bogn> "schema" might be a good search query for you

[15:22:44] <the_voice_> I think the biggest thing that I need to remember is that mongo doesn't have transactions so if you write to multiple docs you may have a problem

[15:23:07] <bogn> not if you have the "multiple" embedded

[15:23:13] <the_voice_> right

[15:23:21] <the_voice_> That I know

[15:23:22] <bogn> then you have atomic operations

[15:23:28] <bogn> rely on them as much as you can

[15:23:47] <bogn> $set, $push and so on

[15:23:52] <the_voice_> and the 16MB file limit doesn't worry me because it would take a lot of comments to get to that :D

[15:23:58] <bogn> yes

[15:25:07] <bogn> but be aware of that, maybe catch an exception or something like that. You might continue comments on a different doc then and post a link on the orignal article or something. And if this happens more that close to never, rethink.

[15:25:27] <bogn> that = than

[15:25:35] <the_voice_> The other concerns I have are Meteor related. Massive updates can cause oplog flooding which can crash the meteor server(although I think they patched that now). But again those are problems for when I scale which I'll deal with then

[15:25:51] <the_voice_> It won't happen. It's for a tool where you won't have more than a 100 comments max on a page

[15:26:29] <bogn> https://www.mongodb.com/presentations/all?utf8=%E2%9C%93&search=Meteor&company_hidden=#

[15:26:40] <bogn> Meteor is also covered

[15:26:46] <the_voice_> read that :)

[15:27:00] <bogn> I don't know if it's just introductions

[15:27:52] <the_voice_> Meteor has a lot of people who seem to use Mongo more like a relational database. I am trying to avoid that right now

[15:30:43] <the_voice_> Thanks a lot bogn, you have been very helpful

[15:31:14] <bogn> you're welocme

[15:31:17] <bogn> welcome

[15:32:34] <bogn> also, if you use atomic operators your oplog is smaller than if you always use full updates

[15:34:54] <the_voice_> oh great

[15:36:21] <bogn> update({name: {$set: "newName"}) instead of update(monsterDoc). The small $set stuff is exactly what's sent to the oplog, same for the monsterDoc.

[15:37:37] <the_voice_> but if you are using subdocuments and you include for instance the username and userId(so you don't have to call the users collection)

[15:37:59] <the_voice_> And then you wanted to update all of those subdocs is it the same thing?

[15:52:59] <bogn> this inclusion of username and userId is called denormalization (some call it caching)

[15:53:23] <bogn> updating multiple sub-docs is something akin to this, if I recall correctly

[15:53:40] <bogn> update(query, {subArray.name: {$set: "newName"})

[15:54:23] <the_voice_> Cool thanks

[15:54:30] <bogn> update(query, {"subArray.name": {$set: "newName"})

[15:54:48] <bogn> might be

[15:54:48] <bogn> update(query, {"subArray.$.name": {$set: "newName"})

[15:55:06] <the_voice_> The problem then is probably meteor having a hard time updating all of it's subscriptions on it's side

[15:55:11] <the_voice_> I think they fixed it though

[15:55:45] <bogn> but this is better than update(query, {subArray: clonedSubArrayWithNewName)

[15:56:28] <bogn> or update(query, {subArray: {$set: clonedSubArrayWithNewName})

[15:56:37] <the_voice_> If it is two deep you can't use the $ though

[15:56:50] <bogn> then you have to use manual indexes

[15:56:51] <the_voice_> Ah right, you are saying don't do two deep

[15:56:54] <the_voice_> right

[15:57:03] <the_voice_> I'll read how mongo recommend I do it

[15:57:10] <bogn> that's my most desired feature

[15:57:26] <bogn> JIRA-831 having this not limited to two-levels

[15:57:32] <bogn> SERVER-831

[15:57:34] <bogn> not JIRA

[15:57:46] <bogn> https://jira.mongodb.org/browse/SERVER-831?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=729172#comment-729172

[15:58:36] <the_voice_> yah I can see it causing issues

[16:04:26] <winem_> hi. is it correct that all databases except for local are replicated in a replicaset?

[16:04:56] <the_voice_> born do I need to worry about storage fragmentation as the document grows? If I understand correctly MongoDB 3.0 has really helped solve that problem

[16:05:00] <the_voice_> *bogn

[16:05:29] <bogn> WiredTiger has

[16:05:56] <bogn> use $push not in-place updates there

[16:06:43] <bogn> moving documents was avoided in mmapv1 with preallocation, not so nice

[16:07:32] <bogn> when appending in WiredTiger you don't run into these issues

[16:07:45] <the_voice_> cool

[16:08:41] <the_voice_> Compose and mongoldb don't let you use it yet :( AH well, will worry about it later :D

[16:08:49] <the_voice_> *mongolab

[16:09:40] <bogn> the switch is rather easy

[16:09:49] <bogn> and saves DB space

[16:09:56] <bogn> through compaction

[16:10:03] <bogn> compression

[16:10:04] <the_voice_> Yah but I host on mongolab for my dev. Will probably put the prod version on Compose.io

[16:10:21] <the_voice_> and they don't have the option yet for wiredtiger

[16:10:38] <bogn> that was to say, you might start with mmapv1 and switch to WiredTiger when documents and users grow

[16:10:45] <the_voice_> oh right

[16:10:55] <mylord> on osx, all my db’s keep disappearing after restart.. what’s that?

[16:11:29] <bogn> mylord: very unusal, never heard that from the Mac people in my office

[16:12:05] <bogn> sorry, can't help there, need to react on customer request now

[16:15:46] <winem_> mylord, did you check that the files in dbpath still exist after shutdown / restart of the mongod?

[16:16:21] <mylord> no, not yet.. where is dbpath usually, on osx?

[16:23:09] <winem_> not sure about osx

[16:23:20] <winem_> do you have an init script or /etc/mongod.conf file?

[16:23:37] <winem_> on linux it's something like /var/lib/mongod/

[16:30:32] <ruairi> mylord: if you're not explicitly setting your dbpath anywhere, it'll look for the files in /data

[16:31:06] <mylord> and on debian?

[16:32:29] <ruairi> That comes with a /etc/mongod.conf if you're using the version in the debian repo

[16:33:06] <ruairi> and the dbpath should be set there.

[16:35:43] <mylord> in /data i see <mydb>.0

[16:35:52] <mylord> and .ns

[16:36:57] <mylord> how would i restore using that?

[16:37:43] <ruairi> you wouldn't

[16:38:02] <ruairi> oh wait

[16:38:13] <ruairi> how do you normally run mongodb

[16:38:14] <ruairi> ?

[16:39:30] <mylord> service mongod start //on debian

[16:39:38] <mylord> on osx i forget how i started but now using brew

[16:40:15] <mylord> on osx, i guess i just ran: mongod

[16:40:19] <mylord> https://docs.mongodb.org/manual/tutorial/install-mongodb-on-os-x/

[16:40:59] <ruairi> Okay, if you're running it on OSX, run mongod --dbpath=/data

[16:45:51] <ruairi> mylord: if you want it to just start on boot on your Mac, try this: https://alicoding.com/how-to-start-mongodb-automatically-when-starting-your-mac-os-x/

[16:46:17] <mylord> looks like that was the path anyhow, hte db files are there

[16:46:26] <mylord> not sure why the db’s all disapperaed tho

[16:46:27] <ruairi> good stuff

[16:46:35] <ruairi> hmm, well

[16:46:36] <mylord> but what about restoring them somehow?

[16:47:11] <ruairi> maybe that wasn't the path. It's possible you were using another path for your dbs, and that you just launched it in /data once and created those files.

[16:47:21] <mylord> anyhow, i think the files in /data got overwritten now with new <mydb>

[16:48:01] <ruairi> they shouldn't have

[16:48:24] <ruairi> if mongod sees existing files, it will open them, rather than overwrite them.

[16:48:52] <ruairi> right, I have to run for the bus, good luck.

[16:50:37] <mylord> ** WARNING: You are running this process as the root user, which is not recommended.

[16:51:52] <mylord> ruairi: i followed instructions on your page there.. did i do somethign wrong?

[16:51:56] <mylord> ps now shows: /usr/local/bin/mongod --dbpath /var/lib/mongodb/ --logpath /var/log/mongodb.log

[17:03:11] <mylord> i’ve previously made a mongodb on my mac and was using that

[17:03:18] <mylord> mongod user

[17:51:33] <Genitrust> i'm looking at this page... https://docs.mongodb.org/manual/tutorial/create-an-auto-incrementing-field/

[17:51:40] <Genitrust> why does an auto increment value "not scale" on mongo?

[17:51:56] <Genitrust> is it because you may easily get more than 4 billion records for what mongo is intended to be used for?

[19:26:35] <jareddlc> hey guys, anyone available to help with mongo connection string?

[19:39:49] <StephenLynx> jareddlc shoot

[19:40:18] <jareddlc> StephenLynx: i would like to add keepalive to the string but i only see two options connectionTimeoutMS and socketTimeoutMS

[19:40:44] <jareddlc> are you familiar with this flag?

[19:40:55] <StephenLynx> nope

[19:41:17] <jareddlc> ok, thanks for helping

[19:41:59] <StephenLynx> v:

[19:42:08] <StephenLynx> btw

[19:42:11] <StephenLynx> what does keepalive does?

[19:42:28] <jareddlc> should allow the conection to stay open

[19:42:35] <StephenLynx> but that is the default

[19:43:10] <jareddlc> you are right

[19:43:14] <jareddlc> it may be a host issue

[19:43:29] <jareddlc> and in that case mongo cant do anything right?

[19:43:40] <StephenLynx> if the connection is faulty, no.

[19:43:56] <StephenLynx> your driver could retry and reconnect thought.

[19:44:08] <StephenLynx> which it does, at least with the node driver.

[19:44:23] <StephenLynx> I even tried restarting the database and the connection didn`t fail on me.

[19:44:40] <StephenLynx> hell, I think I even restarted the db server

[19:44:47] <StephenLynx> and it didn`t complain

[19:44:53] <jareddlc> we recently moved to a different cloud provider

[19:45:05] <jareddlc> and we are having issues after 10 mins of idle the connection drops

[19:45:10] <StephenLynx> cloud providers are usually bad.

[19:45:14] <StephenLynx> which one are you using?

[19:45:28] <jareddlc> well we are using same mongo provider

[19:45:46] <jareddlc> we switch providers for our applications to google compute engine

[19:46:09] <StephenLynx> I am not familiar. I do remember google app engine though.

[19:46:17] <StephenLynx> it is expensive AF from what I remember

[19:46:41] <jareddlc> err container enginge

[19:47:04] <StephenLynx> I would bet my ass they are screwing up.

[19:47:19] <StephenLynx> its just too many layers between the software and the database.

[19:48:06] <jareddlc> well the db is elsewhere not in google container engine

[19:48:16] <jareddlc> anyways thanks for the help

[19:48:18] <StephenLynx> if the network on the software server is failing

[19:48:19] <jareddlc> i'll look into this more

[19:48:29] <StephenLynx> the effect would be the same

[19:48:37] <jareddlc> yup

[19:50:33] <StephenLynx> as a rule of thumb I never use cloud services or containers.

[19:50:39] <StephenLynx> it just makes it harder to debug.

[19:50:44] <StephenLynx> and usually is more expensive

[21:25:22] <kexmex> mongodump is horribly slow latest--- is that a known problem? is there a way to diagnose it?

Log file Viewer

Help | Karma | Search:

#mongodb logs for Friday the 4th of December, 2015