Log file Viewer

[00:00:41] <m3t4lukas> msx: don't know, I always use IRC

[00:01:06] <m3t4lukas> msx: did you try to restart the restoring process?

[00:01:27] <cheeser> msx: there's the mongodb-users google group

[00:01:57] <m3t4lukas> cheeser: who uses google groups :P

[00:02:09] <cheeser> we do for one :)

[00:02:21] <m3t4lukas> yep you do

[00:03:03] <m3t4lukas> but you do good online courses, so that makes it up for it

[00:03:34] <m3t4lukas> I'll go catch some sleep now. See ya all

[00:04:09] <msx> m3t4lukas: yes! I tried with 'mongorestore -v --maintainInsertionOrder --drop --stopOnError DIRECTORY' to no avail :P

[00:04:17] <msx> cheeser: o/ thanks, will check it

[01:28:07] <flaf> Hi @all. Sorry, I'm completely noob concerning mongodb I'm currently discovered. I would like to know if set “noauth=true” is dangerous on a mongo server. Does it mean that all people can open a remote access to mongodb? Correct? But people will not have necessarily access to the bases, correct?

[01:31:09] <cheeser> people who can connect, sure.

[01:31:30] <cheeser> if you lock it down to localhost or a subnet you'd be "fine"

[01:32:32] <flaf> Yes, but in the subnet, a remote user can have access to the admin bases without auth for instance?

[01:32:43] <cheeser> yes

[01:33:23] <flaf> ok, so in production, it's better to have noauth=false, correct?

[01:33:30] <cheeser> absolutely

[01:33:41] <flaf> Ok, cheeser thx for your help.

[01:33:42] <cheeser> always enable auth in production and restrict ip access

[01:35:27] <flaf> ok. To restrict ip access, is there a parameter in the conf or should I use iptables?

[01:36:49] <cheeser> bind to localhost. firewall. etc.

[01:37:06] <flaf> Ok, I see. Thx.

[02:15:10] <shlant> cheeser: is this not possible? echo "rs.slaveOk()" >> ~/.mongorc.js

[02:29:48] <cheeser> shlant: now *that* might work

[02:30:12] <cheeser> secondary reads are a client thing not a server thing so if it's going to work, it'd be there and not mongod.conf

[03:49:33] <dman777_alter> hello, I created a admin user and given the role roles: [ { role: "userAdminAnyDatabase", db: "admin" } ] in db called admin. when I 'use curves' db, this user won't authenticate and can not do operations. Why, since it has userAdminAnyDatabase?

[04:07:09] <joannac> dman777_alter: erm, "userAdmin" means "can create, delete, and modify users"

[04:07:30] <joannac> the user is in the admin database, so you need to authenticate with it in the admin database

[04:12:00] <dman777_alter> joannac: I did...and then I went did 'use curves' and tried to create a collection and I wasn't allowed to

[04:12:59] <dman777_alter> also, I am not allowed to do listCollections on curves: "errmsg" : "not authorized on curves to execute command { listCollections: 1.0, filter: {} }"

[04:19:09] <joannac> dman777_alter: do you think listing collections falls under "can create, delete, and modify users"

[04:20:23] <dman777_alter> joannac: ah...ok. makes sense. I thought userAdminAnyDatabase would mean also full roles on everything.

[04:21:15] <dman777_alter> joannac: after enabling access control, is there a such root user with full access?

[04:25:32] <joannac> dman777_alter: not really. "root" is close

[04:28:53] <dman777_alter> joannac: what is the benifit/difference of useing metadata versus a field?

[04:34:20] <Boomtime> dman777_alter: "metadata versus a field" <- can you clarify what you mean by this?

[04:37:21] <dman777_alter> Boomtime: well, I am makeing a document that will have user information such as username, location, date created, birthday. I will also have 10 fields available for names of thier pets. I wasn't sure if they should be metadata or not. I don't really know the difference between a metadata block of fields and normal fields

[04:40:21] <Boomtime> dman777_alter: json has no such defined difference so i'm still not sure what you're referring to - can you provide an example of both?

[04:40:54] <dman777_alter> Boomtime: well, I was wondering what is the use case of metadata

[04:41:12] <Boomtime> and what is metadata in your understanding? again, json has no such thng

[04:41:48] <dman777_alter> I'm reading https://docs.mongodb.org/ecosystem/use-cases/metadata-and-asset-management/ right now. To me, metadata is details information about a key. But that could be done without metadata

[04:42:26] <dman777_alter> I could just use nested key/properties on a key for that without metadata

[04:42:31] <Boomtime> metadata is a conceptual difference in that example

[04:43:14] <Boomtime> there is literally a field named "metadata"

[04:43:32] <Boomtime> what if that field was named "properties" instead? would anything change?

[04:44:04] <Boomtime> or we could have named it "details"

[04:44:08] <Boomtime> etc.

[04:48:37] <dman777_alter> ok...cool. so no special meaning to mongodb then

[04:49:32] <Boomtime> mongodb doesn't enter in to it - metadata is just a label you give something, like calling something a 'property', or a 'detail'... it says nothing about how you store that information

[04:49:51] <dman777_alter> by best practices, when would I use the label metadata?

[04:50:22] <Boomtime> the dictionary definition of 'metadata' is "data about data"

[04:50:48] <dman777_alter> ok, I assume then it would be used to store data about the document itself?

[04:52:04] <Boomtime> take an example of a filesystem, you might refer to the created date and last modified timestamps as 'metadata' about the file

[04:52:22] <Boomtime> again, it's just a conceptual model, it doesn't tell you anything about the data itself

[04:52:35] <Boomtime> nor the format

[04:52:40] <dman777_alter> Boomtime: cool, thanks

[04:52:48] <Boomtime> which i'm beginning to suspect is what you really want to know

[04:53:06] <Boomtime> i.e what format for your document should you use - is that what you mean?

[04:53:19] <dman777_alter> yep

[04:53:47] <Boomtime> for example, a document using fields directly: { name: "bebop", age: 28, location: "paraguay" } versus a something like a properties array: { properties: [ {k:"name",v:"bebop"}, {k:"age", v:28 }... ] }

[04:54:49] <dman777_alter> ya, I didn't even take an array into account. For a user profile with pets.... which would be correct format to have the pets in the users document?

[04:55:53] <Boomtime> that would be very use case dependent - perhaps? - how many pets do you allow? if the answer is 'infinite' then you probably don't want to store them as an embedded array since a single document has an upper limit on size

[04:56:24] <Boomtime> if the answer is, say, 'less than a thousand'.. then embedded array in the user document would probably be fine

[04:58:39] <dman777_alter> Boomtime: I was thinking a limit of 10 pets

[04:58:55] <Boomtime> right, then embedded array makes sense

[04:59:04] <dman777_alter> Boomtime: cool....thanks!

[09:08:32] <flaf> Hi, I have 3 servers mongo. It seems to me that the replicaset is ok. For instance, if I stop srv-1, I can see after few seconds PRIMARY in the prompt fo srv-2. But is there a command to see if each server is up and ok?

[09:17:29] <m3t4lukas> msx: did you resolve your restoration problem?

[09:53:33] <profsimm> Hey folks am I supposed to have BOTH mongod.conf and mongodb.conf in my init.d?

[09:53:43] <profsimm> Because when I manually start the mongodb service it works

[09:53:50] <profsimm> But when I reboot, it doesn't start on its own

[09:58:38] <profsimm> Folks!

[09:58:39] <profsimm> :P

[12:23:57] <m3t4lukas> profsimm: sry, was afk

[12:24:28] <m3t4lukas> oh, he's gone

[12:24:46] <m3t4lukas> anyway profsimm if you read the logs: it's an OS thing

[21:42:11] <monkh> Hi, I followed the mongoose tutorial on their webpage and it appears to be working but when i try find the kittens collection in robomongo or mongo cli i can see test db but shows no collections inside it. how do i reveal my schrodingers kittens in cli/robomongo?

[21:43:03] <monkh> http://pastebin.com/eYtsQtQv

[21:43:06] <monkh> here is a copy of my js

[21:59:35] <msx> m3t4lukas: hey, good afternoon! (utc -3 here). I wasn't able yet, I left a post at Percona's forums (we are using their stack) with all the information I have so far: https://www.percona.com/forums/questions-discussions/percona-server-for-mongodb/43561-mongorestore-hangs-restoring-a-mongodb-2-6-dump-to-a-ps-for-mongodb-3-0-8-wiredtiger

[22:02:09] <msx> I will try later to find the solution though, it would be nice to start the week with this issue solved

[22:02:52] <m3t4lukas> msx: i can imagine

[22:04:06] <m3t4lukas> msx: have you ever tried PerconaFT storage engine?

[22:38:07] <msx> m3t4lukas: not yet, I'm working with wiredTiger because that's the one my boss and the dev team decided to use

[23:04:07] <sandstrom> I need to support two queries with indices: (1) on keys `a` and `b`, and (2) on keys `b` and `c`. Should I use two separate compound indices, one per query, or a single compound index covering a/b/c?

[23:21:44] <metame> sandstrom: depends on how you're using your data, specifically how you're sorting it

[23:22:05] <sandstrom> (without sorting)

[23:22:54] <metame> most likely two separate indices would be more performant

[23:25:22] <metame> a single compound index wouldn't make sense

[23:25:44] <metame> bc a query on an index always needs to include the leftmost thing

[23:26:12] <metame> if your 2nd query required a as well then a compound index could work of a/b/c

[23:26:30] <metame> and the index always sorts by the first key btw

[23:27:16] <metame> so for your example two compound indexes (a/b & b/c) would be the most performant

[23:28:13] <metame> and give you fully "covered" queries (completely satisfied by an index)

[23:28:43] <metame> also recommend filtering out _id key if that's what you're going fo

[23:28:45] <metame> *for

[23:51:45] <metame> sandstrom: that take care of it?

[23:54:15] <sandstrom> metame thanks!

[23:54:25] <sandstrom> metame the last part, filtering out _id, can you elaborate?

[23:57:01] <metame> sure. so _id is by default included on every query

[23:58:02] <metame> and you want to project it out in this case

[23:58:34] <metame> you can project it out any time you don't actually need it but one of the main times you'd do that is to get a fully covered query by an indexx