| openstackgerrit | MD NADEEM proposed openstack/python-zaqarclient: Attaching a default project_id in request header https://review.openstack.org/243969 | 00:04 |
|---|---|---|
| *** achanda has joined #openstack-zaqar | 00:15 | |
| *** achanda has quit IRC | 00:35 | |
| Eva-i | Hey, everyone. Do you think zaqar's tox need py26 and py33 environments? | 00:46 |
| flwang | Eva-i: openstack has drop the support for py26 | 00:46 |
| Eva-i | flwang1: aha. According to information on https://wiki.openstack.org/wiki/Python3 support for python 3.3 is dropped too. | 00:47 |
| Eva-i | Alright | 00:47 |
| flwang | cool | 00:49 |
| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Fix tox not performing some tests in some Envs https://review.openstack.org/242182 | 01:01 |
| mdnadeem | flwang: Ping | 01:01 |
| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Fix tox not performing some tests in some Envs https://review.openstack.org/242182 | 01:03 |
| Eva-i | flwang: is it alright that I cleaned py26 and py33 environments in the same patch? | 01:04 |
| flwang | yep, it's fine and i can nijia approve it, so don't worry about the current +2 | 01:08 |
| *** szaher has quit IRC | 01:11 | |
| Eva-i | Okay. I heard openstack encourages developers to split changes as much as possible. It's like it doesn't support "boy scout rule" which I sometimes find being used in other projects. | 01:14 |
| *** kgriffs is now known as kgriffs|afk | 01:16 | |
| flwang | this one doesn't apply for the 'split' | 01:17 |
| *** mdnadeem has quit IRC | 01:21 | |
| *** mdnadeem has joined #openstack-zaqar | 01:21 | |
| *** szaher has joined #openstack-zaqar | 01:23 | |
| mdnadeem | flwang: Hi | 01:23 |
| flwang1 | mdnadeem: hey | 01:23 |
| mdnadeem | flwang: Need to discuss about UnauthorizedError throw by zaqar client | 01:24 |
| mdnadeem | Have a look at the code line : https://github.com/openstack/python-zaqarclient/blob/master/zaqarclient/auth/__init__.py#L26 | 01:24 |
| flwang1 | yes | 01:26 |
| mdnadeem | Default authentication at client side set to "noauth" and server side it is "keystone" | 01:26 |
| mdnadeem | So i am worried how to set "noauth" at server side? | 01:26 |
| mdnadeem | This is the main cause of UnauthorizedError threw by zaqar server when client wants to communicate with it. | 01:28 |
| flwang1 | mdnadeem: i noticed the problem and i'm working on the patch | 01:29 |
| flwang1 | personally, i would like to use keystone as the default auth | 01:30 |
| mdnadeem | yes, that make sense | 01:30 |
| mdnadeem | flwang1: i Have write some code to fix that would you like to see that? | 01:32 |
| flwang1 | mdnadeem: yep, sure | 01:33 |
| mdnadeem | flwang1: give me a moment | 01:34 |
| flwang1 | ok | 01:34 |
| *** itisha has joined #openstack-zaqar | 01:39 | |
| mdnadeem | flwang1: http://paste.openstack.org/show/478622/ | 01:40 |
| mdnadeem | flwang1: more focus on line 19-20 , this is the point where get authentication parameters for keystone authentication | 01:42 |
| flwang1 | mdnadeem: that looks fine, but we don't need the change zaqarclient/auth/keystone.py | 01:43 |
| flwang1 | besides, i'm worrying about if this can pass the test | 01:43 |
| mdnadeem | flwang1: In my Environment , i have check it works fine | 01:44 |
| flwang1 | mdnadeem: cool | 01:44 |
| flwang1 | so why we need the change for zaqarclient/auth/keystone.py ? | 01:45 |
| mdnadeem | at line 19-20 , We fetch authentication parameters, it is in the form of "os_username" and not in the form of "--os_username" | 01:47 |
| mdnadeem | sorry | 01:48 |
| mdnadeem | "username" not in the form on "os_username" | 01:48 |
| mdnadeem | flwang1: did you get my point? | 01:48 |
| flwang1 | mdnadeem: not really, so where are you getting the username then? | 01:49 |
| mdnadeem | at line 19-20, instance._auth_params get value like {'username':"admin", "password":'pass'....} | 01:52 |
| mdnadeem | however on keystone.py we try access the value using "os-username" | 01:54 |
| mdnadeem | Thats why i make change in keystone.py and change "os-username" to "username" and same with other parameter too. | 01:55 |
| flwang1 | i don't understand actually, you didn't change the key of dict, but the value, so why it's related to the code at zaqarclient/queues/cli.py | 01:57 |
| flwang1 | Eva-i: still around? | 01:58 |
| mdnadeem | flwang1: Well , the cause to change the keystone.py is that at line 30, we access value of the self.conf by self.conf.get('os_username'), however self.conf has no key 'os_username' instead it has key 'username' | 02:06 |
| Eva-i | flwang1: yes, i'm here | 02:10 |
| flwang1 | Eva-i: i tried the wc issue, it doesn't work | 02:10 |
| flwang1 | we can't set the wc after initialized the mongoclient | 02:10 |
| flwang1 | mdnadeem: i'm testing your way | 02:11 |
| flwang1 | to understand it better | 02:11 |
| mdnadeem | flwang1: Thanks :) | 02:11 |
| Eva-i | flwang1: you mean we can't do this https://github.com/openstack/zaqar/blob/master/zaqar/storage/mongodb/driver.py#L167 if we initialize mongoclient with wc? | 02:12 |
| flwang1 | no | 02:12 |
| flwang1 | Eva-i: i mean we can't initialize it when creating the mongoclient | 02:12 |
| flwang1 | and you know the error is raising before accessing the database, but the connection | 02:13 |
| Eva-i | flwang1: yes, I know, because "connection" is created before any interactions with databases | 02:16 |
| flwang1 | Eva-i: yes | 02:16 |
| Eva-i | flwang1: that's strange, that you wasn't able to initialize mongoclient with wc | 02:17 |
| Eva-i | flwang1: i'll try it too | 02:17 |
| flwang1 | and mongoclient will raise an exception if we just set the default value for w not matter what's the mongodb version :( | 02:17 |
| flwang1 | Eva-i: we can for mongodb version >=2.6 | 02:17 |
| flwang1 | my mongodb is 2.4.9 the default version on ubuntu 14.04 | 02:18 |
| flwang1 | so i can see the error | 02:18 |
| flwang1 | it would be great if you can try it on your local | 02:18 |
| Eva-i | okay, I'll try now | 02:19 |
| flwang1 | mdnadeem: you can try again without the change for zaqarclient/auth/keystone.py, i think you will get the same result | 02:21 |
| mdnadeem | flwang1: ohk, i ll try it :) | 02:23 |
| flwang1 | mdnadeem: i can see your point now | 02:40 |
| mdnadeem | flwang1: So what u get the vale of instance._auth_params? | 02:41 |
| flwang1 | mdnadeem: i don't like current way, TBH. but i haven't figured out a better way :) | 02:41 |
| flwang1 | besides, we should use token first | 02:42 |
| flwang1 | to avoid talk with keystone | 02:42 |
| mdnadeem | ohk, | 02:42 |
| mdnadeem | flwang1: did you mean Token authentication method in place of password authentication method ? | 02:43 |
| mdnadeem | http://docs.openstack.org/developer/python-keystoneclient/authentication-plugins.html#v2-identity-plugins | 02:45 |
| Eva-i | flwang1: what exactly exception it raises? | 02:47 |
| flwang1 | i mean openstact client has already got the token | 02:47 |
| flwang1 | so we should use that | 02:48 |
| flwang1 | Eva-i: it has been flushed | 02:48 |
| flwang1 | but i'm sure there is an exception | 02:48 |
| flwang1 | how about your test? | 02:48 |
| flwang1 | mdnadeem: besides, we also need to set the value for 'insecure' and 'cacert' | 02:52 |
| Eva-i | flwang1: I don't see any exception when for testing purposes I intantiate MongoClient like this inside _connection() method: | 02:53 |
| Eva-i | http://paste.openstack.org/show/UpFNGi4RQXmKxVeqULNv/ | 02:53 |
| flwang1 | Eva-i: ok, so two questions | 02:53 |
| Eva-i | zaqar-server runs smoothly, creates indexes and adding records in mongodb. I'm on ubuntu 14.04 too | 02:53 |
| flwang1 | 1. what's your mongodb version | 02:53 |
| flwang1 | 2. are you sure the code is executed? | 02:54 |
| *** achanda has joined #openstack-zaqar | 02:54 | |
| Eva-i | 1. db version v2.4.9 | 02:54 |
| Eva-i | 2. sure | 02:54 |
| Eva-i | I just add this code after line 41 inside _connection() method | 02:55 |
| Eva-i | hmm... | 02:56 |
| Eva-i | 2. let me overcheck | 02:56 |
| flwang1 | you can just raise an exception to make sure it's executed | 02:57 |
| mdnadeem | flwang1: hmm, it sound good if we can re use the token , it will reduce overhead to connect with keystone :) | 02:58 |
| flwang1 | mdnadeem: yes | 02:58 |
| flwang1 | so feel free upload a new patch set | 02:58 |
| flwang1 | patch i mean | 02:58 |
| mdnadeem | flwang1: Yah sure, i would like to work on it | 02:59 |
| flwang1 | we need to get it in asap, and think it's breaking the use of zaqar client | 02:59 |
| mdnadeem | yes, i will try asap | 02:59 |
| mdnadeem | :) | 02:59 |
| Eva-i | flwang1: I raised and noticed that no exception was thrown, zaqar from other repo was installed on system | 02:59 |
| flwang1 | Eva-i: haha | 03:00 |
| Eva-i | flwang1: yeah -_- | 03:00 |
| flwang1 | anyway, i will upload a new patch set and give it a clear note | 03:00 |
| flwang1 | feel free to comment it based on your test | 03:01 |
| flwang1 | mdnadeem: thanks | 03:01 |
| flwang1 | mdnadeem: let me know if you need any help | 03:01 |
| mdnadeem | sure | 03:02 |
| flwang1 | mdnadeem: not sure if you notice the problem, even though there is a 401 error, but the command line doesn't print anything | 03:03 |
| Eva-i | flwang1: 2. it fails with "norepl" error as flaper87 stated here: https://github.com/openstack/zaqar/blob/master/zaqar/storage/mongodb/driver.py#L153 | 03:03 |
| Eva-i | maybe you could reuse this version check condition when instantiating MongoClient | 03:03 |
| *** szaher has quit IRC | 03:04 | |
| flwang1 | Eva-i: not really | 03:05 |
| flwang1 | oh, yep norepl | 03:05 |
| flwang1 | but we can't reuse the check | 03:05 |
| flwang1 | since we have to create the connection to check the version | 03:10 |
| flwang1 | and it doesn't make sense to create the connection again to set the wc value | 03:10 |
| mdnadeem | flwang1: yes i had notice it earlier. However When i run command using --debug option then it shows unauthorization error. Thats y i dnt take it seriously | 03:14 |
| mdnadeem | flwang1: But yes it should through exception with out --debug option too. I will check the code | 03:15 |
| mdnadeem | flwang1: Thanks for your input :) | 03:15 |
| Eva-i | flwang1: right. If I'm not mistaken connection is mutable and can be changed. What if after gathering connection info as usual in __init__(), we check it and modify connection's write concern if needed | 03:16 |
| Eva-i | *gathering connection's servers version | 03:16 |
| *** szaher has joined #openstack-zaqar | 03:16 | |
| flwang1 | Eva-i: seems there is not way to change the connection after created | 03:16 |
| flwang1 | at least for the wc | 03:16 |
| flwang1 | mdnadeem: it deserves another bug, i will open it | 03:17 |
| mdnadeem | flwang1: good :) | 03:17 |
| flwang1 | mdnadeem: let's rock on ;) | 03:17 |
| mdnadeem | hahaha | 03:18 |
| Eva-i | flwang1: hm, no way? But I see something like this happens here: https://github.com/openstack/zaqar/blob/master/zaqar/storage/mongodb/driver.py#L68 | 03:19 |
| flwang1 | i don't understand | 03:20 |
| flwang1 | line 68 is just initializing the class | 03:20 |
| flwang1 | my point is you have to create the connection and then check the version | 03:20 |
| flwang1 | and then, initializing the connection again to set the wc value | 03:21 |
| flwang1 | it doesn't make sense for me | 03:21 |
| Eva-i | flwang1: it is not, instantiation happens on line 43 | 03:21 |
| flwang1 | line 43 is not instantiation | 03:22 |
| flwang1 | it's deciding which class will be used | 03:22 |
| Eva-i | oh... | 03:22 |
| flwang1 | does that make more sense? | 03:23 |
| Eva-i | flwang1: yes, I was wrong | 03:23 |
| flwang1 | glad to see we're on the same page now :) | 03:24 |
| Eva-i | flwang1: create one connection just to check mongodb's version and then create another connection with right write concern parameter may seem a strange, but working solution | 03:30 |
| flwang1 | i agree, but i don't like that way | 03:31 |
| flwang1 | the effort is too big, IMO | 03:31 |
| Eva-i | flwang1: I think it will require just adding few lines of code | 03:32 |
| flwang1 | not related to lines of code | 03:33 |
| flwang1 | but the performance | 03:33 |
| Eva-i | ok | 03:33 |
| Eva-i | I'm not sure, but I think databases are not being created too often | 03:36 |
| Eva-i | *connections | 03:36 |
| flwang1 | maybe, but the benefit is not bigger than the effort compared with the other way. and again, that way is too ugly, IMHO | 03:39 |
| openstackgerrit | Fei Long Wang proposed openstack/python-zaqarclient: Remove periods of docstring https://review.openstack.org/244421 | 03:42 |
| Eva-i | don't forget there's also method №1 - "create all databases by using connection.get_database() with passing write concern". There are not many places in zaqar/storage/mongodb/driver.py where databases are created. | 03:42 |
| Eva-i | maybe we can just use current patch for now and open separate bug report for "write concern" issues. | 03:44 |
| flwang | 'There are not many places in zaqar/storage/mongodb/driver.py where databases are created' i didn't get your point | 03:46 |
| Eva-i | flwang1: i just mean it would be not hard to fix all these places where databases are created in the code | 03:47 |
| Eva-i | flwang1: ok, it was nice to talk with you tonight, but I suspect I wasn't really helpful to you... instead you was helpful by enlightening me =) | 03:47 |
| flwang | i'm not an expert of mongodb, but based on my understanding | 03:48 |
| flwang | wc is a high cost action or 'slow' action | 03:48 |
| flwang | my point is for message, we would like to keep the durability | 03:49 |
| flwang | so we need it, but i'm not sure if we really need it for queue and subscription | 03:49 |
| flwang | we need more discussion about that | 03:49 |
| flwang | it would be nice if you can open a bug to track it | 03:49 |
| *** itisha has quit IRC | 03:51 | |
| Eva-i | flwang1: a bug report where I ask queue and subscription databases must be durable or not? | 03:54 |
| Eva-i | *ask if | 03:54 |
| flwang | sort of | 03:54 |
| flwang | i have to run now | 03:54 |
| Eva-i | flwang: okay | 03:54 |
| flwang | ttyl, see you guys | 03:54 |
| flwang | Eva-i: thank you so much for your effort on this | 03:54 |
| Eva-i | flwang: oh, thank you too | 03:55 |
| *** flwang1 has quit IRC | 03:57 | |
| openstackgerrit | MD NADEEM proposed openstack/python-zaqarclient: Added cli implementation for v2 https://review.openstack.org/241523 | 04:22 |
| *** khushbu_ has joined #openstack-zaqar | 04:29 | |
| openstackgerrit | MD NADEEM proposed openstack/python-zaqarclient: Implement CLI support for pool delete https://review.openstack.org/242388 | 04:30 |
| *** achanda has quit IRC | 05:04 | |
| *** achanda has joined #openstack-zaqar | 06:05 | |
| *** achanda has quit IRC | 06:10 | |
| *** achanda has joined #openstack-zaqar | 06:15 | |
| *** ryansb_ has joined #openstack-zaqar | 06:26 | |
| *** ryansb_ has quit IRC | 06:26 | |
| *** ryansb_ has joined #openstack-zaqar | 06:26 | |
| *** ryansb has quit IRC | 06:26 | |
| *** ryansb_ is now known as ryansb | 06:26 | |
| *** khushbu_ has quit IRC | 07:07 | |
| *** elynn has joined #openstack-zaqar | 07:18 | |
| *** achanda has quit IRC | 07:20 | |
| *** elynn has left #openstack-zaqar | 07:25 | |
| *** khushbu_ has joined #openstack-zaqar | 07:58 | |
| *** akanksha_ has joined #openstack-zaqar | 08:22 | |
| *** dynarro has joined #openstack-zaqar | 08:26 | |
| *** Pablo|off| has joined #openstack-zaqar | 08:40 | |
| *** achanda has joined #openstack-zaqar | 08:47 | |
| *** Pablo|off| is now known as pcaruana | 08:48 | |
| *** rcernin has joined #openstack-zaqar | 08:50 | |
| *** AAzza has quit IRC | 09:12 | |
| *** AAzza has joined #openstack-zaqar | 09:12 | |
| *** AAzza has quit IRC | 09:21 | |
| *** AAzza has joined #openstack-zaqar | 09:25 | |
| *** achanda has quit IRC | 09:34 | |
| *** openstackgerrit has quit IRC | 09:46 | |
| *** openstackgerrit has joined #openstack-zaqar | 09:46 | |
| *** khushbu_ has quit IRC | 09:49 | |
| *** openstackgerrit has quit IRC | 10:16 | |
| *** openstackgerrit has joined #openstack-zaqar | 10:16 | |
| *** dynarro has quit IRC | 10:21 | |
| *** akanksha_ has quit IRC | 10:38 | |
| *** rcernin_ has joined #openstack-zaqar | 10:48 | |
| *** khushbu_ has joined #openstack-zaqar | 10:48 | |
| *** rcernin has quit IRC | 10:49 | |
| *** rcernin has joined #openstack-zaqar | 10:50 | |
| *** rcernin_ has quit IRC | 10:53 | |
| *** khushbu_ has quit IRC | 10:58 | |
| *** khushbu_ has joined #openstack-zaqar | 11:01 | |
| *** dmowrer has joined #openstack-zaqar | 12:14 | |
| *** AAzza has quit IRC | 12:51 | |
| *** AAzza_ has joined #openstack-zaqar | 12:51 | |
| *** AAzza has joined #openstack-zaqar | 12:55 | |
| *** AAzza_ has quit IRC | 12:56 | |
| *** boris-42 has quit IRC | 13:48 | |
| *** AAzza_ has joined #openstack-zaqar | 13:51 | |
| *** AAzza has quit IRC | 13:52 | |
| *** AAzza_ is now known as AAzza | 13:52 | |
| *** AAzza has quit IRC | 13:57 | |
| *** dmowrer has quit IRC | 13:59 | |
| *** AAzza has joined #openstack-zaqar | 14:01 | |
| *** amitgandhinz has joined #openstack-zaqar | 14:13 | |
| *** ametts has quit IRC | 14:24 | |
| *** dmowrer has joined #openstack-zaqar | 14:24 | |
| *** sriram has joined #openstack-zaqar | 14:25 | |
| *** kgriffs|afk is now known as kgriffs | 14:35 | |
| *** khushbu_ has quit IRC | 14:50 | |
| jasondotstar | hi all o/ | 15:06 |
| *** ametts has joined #openstack-zaqar | 15:49 | |
| *** achanda has joined #openstack-zaqar | 15:51 | |
| *** dynarro has joined #openstack-zaqar | 15:56 | |
| *** achanda has quit IRC | 16:02 | |
| *** csoukup has joined #openstack-zaqar | 16:02 | |
| *** achanda has joined #openstack-zaqar | 16:03 | |
| *** achanda has quit IRC | 16:04 | |
| jasondotstar | zigo: ping | 16:11 |
| jasondotstar | zigo: you around? If not, I'll try back later. Re: pkgs for debian jessie | 16:13 |
| *** davideagnello has quit IRC | 16:17 | |
| *** davideagnello has joined #openstack-zaqar | 16:17 | |
| *** rcernin has quit IRC | 16:50 | |
| *** sriram has quit IRC | 17:04 | |
| *** pcaruana has quit IRC | 17:06 | |
| *** exploreshaifali has joined #openstack-zaqar | 17:13 | |
| exploreshaifali | Hello everyone! | 17:15 |
| jasondotstar | exploreshaifali: howdy. chan has been quiet today | 17:18 |
| exploreshaifali | jasondotstar, yea! but we are making little noise | 17:19 |
| exploreshaifali | ;) | 17:19 |
| jasondotstar | lol | 17:19 |
| jasondotstar | +1 | 17:19 |
| exploreshaifali | :D | 17:19 |
| exploreshaifali | jasondotstar, how is your puppet patch? | 17:19 |
| *** flwang1 has joined #openstack-zaqar | 17:23 | |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar: Fix write concern check for py34 https://review.openstack.org/242287 | 17:23 |
| *** rcernin has joined #openstack-zaqar | 17:24 | |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar: Fix write concern check for py34 https://review.openstack.org/242287 | 17:25 |
| *** dmowrer has quit IRC | 17:28 | |
| *** dmowrer has joined #openstack-zaqar | 17:43 | |
| *** dynarro has quit IRC | 17:44 | |
| *** dmowrer has quit IRC | 17:46 | |
| *** dmowrer has joined #openstack-zaqar | 17:46 | |
| *** david-lyle has quit IRC | 17:46 | |
| jasondotstar | exploreshaifali: making progress. need to talk to zigo abt the debian pkgs | 17:47 |
| exploreshaifali | Cool! | 17:47 |
| Eva-i | Hello | 17:55 |
| Eva-i | Has someone tested Zaqar when it's configured to use Mongos or Replica Set? | 17:56 |
| *** stevelle has left #openstack-zaqar | 17:57 | |
| Eva-i | I mean when link provided by user in zaqar.conf is actually pointing to Mongos or MongoDB replica set. | 17:58 |
| Eva-i | *the link | 17:59 |
| Eva-i | Anyone? | 18:01 |
| Eva-i | flwang: I'm writing the bug report I promised to make. When it will be ready, you may include the link to this bug report directly as comment in the code in your patch. | 18:03 |
| *** sriram has joined #openstack-zaqar | 18:08 | |
| Eva-i | flwang1: flwang: are you here? | 18:21 |
| *** achanda has joined #openstack-zaqar | 18:26 | |
| *** flwang1 has quit IRC | 18:28 | |
| zigo | jasondotstar: Go ahead, may I help? | 18:39 |
| jasondotstar | hey zigo | 18:41 |
| jasondotstar | so you mentioned that there was some sort of issue with the debian pkgs for jessie | 18:42 |
| jasondotstar | were you able to work thru any of that? | 18:42 |
| *** EmilienM has quit IRC | 18:42 | |
| *** EmilienM has joined #openstack-zaqar | 18:46 | |
| zigo | jasondotstar: I haven't said that. I wrote I didn't functional test the package and I expect some feedback. | 18:50 |
| jasondotstar | zigo: ack. | 18:51 |
| jasondotstar | zigo: any ETA? | 18:52 |
| *** kgriffs is now known as kgriffs|afk | 18:56 | |
| *** kgriffs|afk is now known as kgriffs | 18:58 | |
| *** kgriffs is now known as kgriffs|afk | 19:01 | |
| *** kgriffs|afk is now known as kgriffs | 19:02 | |
| *** Qiming has joined #openstack-zaqar | 19:10 | |
| *** sriram has quit IRC | 19:12 | |
| *** Qiming_ has joined #openstack-zaqar | 19:14 | |
| *** Qiming has quit IRC | 19:17 | |
| *** achanda has quit IRC | 19:21 | |
| *** achanda has joined #openstack-zaqar | 19:21 | |
| *** csoukup has quit IRC | 19:23 | |
| *** Qiming_ has quit IRC | 19:25 | |
| *** flwang1 has joined #openstack-zaqar | 19:37 | |
| *** csoukup has joined #openstack-zaqar | 19:41 | |
| Eva-i | Can someone confirm(or make invalid) this bug: https://bugs.launchpad.net/zaqar/+bug/1515737 =) | 19:47 |
| openstack | Launchpad bug 1515737 in zaqar ""Reliable" Zaqar will not start when configured to use MongoDB replica set or Mongos" [Undecided,New] - Assigned to Eva Balycheva (ubershy) | 19:47 |
| flwang1 | Eva-i: i'm reading | 19:50 |
| *** AAzza has quit IRC | 19:50 | |
| *** AAzza has joined #openstack-zaqar | 19:50 | |
| Eva-i | flwang1: thanks, Fei Long | 19:51 |
| jasondotstar | zigo: right now, I'm pointing the puppet module to http://liberty-jessie.pkgs.mirantis.com/debian jessie-kilo-backports main | 19:52 |
| jasondotstar | is this still valid? | 19:52 |
| jasondotstar | hmm perhaps not | 19:53 |
| jasondotstar | liberty now i guess... | 19:53 |
| flwang1 | Eva-i: so you mean you setup a replicaset, and still got the exception, right? | 19:59 |
| zigo | jasondotstar: Yes, it is. | 20:02 |
| zigo | liberty, right | 20:02 |
| zigo | Not kilo. | 20:02 |
| *** AAzza has quit IRC | 20:03 | |
| *** AAzza has joined #openstack-zaqar | 20:04 | |
| jasondotstar | zigo: ack. | 20:05 |
| Eva-i | flwang1: I setup mongos instance and added 1 mongodb instance to it, just to make it work. Then in zaqar.conf I modified mongodb url so it points to my mongos instance. Deleted "unreliable = True" parameter from zaqar.conf. Started zaqar-server. | 20:05 |
| jasondotstar | fixing it up now... | 20:05 |
| Eva-i | flwang1: I haven't setup mongodb replica set, but apparently it will not work too. | 20:06 |
| *** Qiming_ has joined #openstack-zaqar | 20:06 | |
| Eva-i | flwang1: exception is different. It's another bug, not related to python's version. | 20:08 |
| flwang1 | python's version? i'm confused | 20:09 |
| flwang1 | you mean the bug i'm fixing? | 20:09 |
| Eva-i | Yes, the bug you're fixing is directly related to python's version. | 20:10 |
| Eva-i | "and still got the exception" I thought you're asking if I still have the same TypeError exception. | 20:10 |
| Eva-i | Exception I mentioned in my bug report is now RuntimeError. | 20:11 |
| *** Qiming_ has quit IRC | 20:12 | |
| flwang1 | i mean if you can still see the same runtimeerror when using replicaset | 20:12 |
| flwang1 | Eva-i: and can you show me your zaqar.conf? | 20:13 |
| Eva-i | flwang1: RuntimeError will be thrown in either zaqar is connected to mongos or to replica set, i.e. when this condition is false: "if not len(conn.nodes) > 1 and not conn.is_mongos:" | 20:14 |
| Eva-i | flwang1: sure | 20:15 |
| Eva-i | flwang1: http://paste.openstack.org/show/1HmdtabmInsN9xgN8ycL/ my mongos is listening to port 27020. | 20:17 |
| flwang1 | Eva-i: so you said you have setup the mongos, right? | 20:18 |
| Eva-i | flwang1: right | 20:18 |
| flwang1 | but you still only have 1 node, right? | 20:18 |
| Eva-i | flwang1: oh... yes | 20:18 |
| flwang1 | 1 node is not reliable | 20:18 |
| Eva-i | flwang1: let me check | 20:19 |
| flwang1 | mongos is just a 'router'. it doesn't store any data | 20:20 |
| flwang1 | so even you have setup mongos, but if there is only 1 node, zaqar still believes it's unreliable, does that make sense? | 20:21 |
| ryansb | Eva-i: yeah, mongos is a "router" that will send requests to different shards or different members of a replica set | 20:22 |
| ryansb | mongod stores data, and you need >1 of them to be "reliable" | 20:23 |
| flwang1 | ryansb: https://review.openstack.org/#/c/242287/ just addressed your comments, welcome to revisit :) | 20:26 |
| ryansb | yup, already had it open to review | 20:26 |
| flwang1 | ryansb: Eva-i and i just give much love the the mongodb driver :) | 20:27 |
| ryansb | :) | 20:28 |
| Eva-i | flwang1: ryansb: tested now with two mongodb instances added as shards to mongos instance: http://paste.openstack.org/show/4IMcJQIiwp2XOuDMOJpI/ . The error is still the same: http://paste.openstack.org/show/hk0hHYcVp6BUBjaKc3f2/ | 20:38 |
| Eva-i | maybe I don't understand something | 20:38 |
| Eva-i | router you say... | 20:39 |
| flwang1 | Eva-i: can you add a breakpoint at line if not len(conn.nodes) > 1 and not conn.is_mongos: | 20:40 |
| *** csoukup has quit IRC | 20:40 | |
| flwang1 | to check how many nodes from conn and what's the is_mongos value | 20:40 |
| Eva-i | flwang1: okay. So "len(conn.nodes)" = 1, "conn.is_mongos" = True | 20:44 |
| Eva-i | flwang1: so the program goes to "else" block and throw the exception | 20:46 |
| flwang1 | that's why | 20:46 |
| flwang1 | only 1 node | 20:47 |
| ryansb | evayeah, that's a correct error though | 20:47 |
| ryansb | shards don't provide durability | 20:47 |
| ryansb | because data on one is not a copy, it's a different "slice" of the database | 20:47 |
| Eva-i | flwang1: ryansb: the program will go to "else" block even if "len(conn.nodes)" will be 2 or more | 20:48 |
| flwang1 | Eva-i: how? | 20:49 |
| ryansb | because it's connected via mongos | 20:49 |
| ryansb | not directly to mongod | 20:49 |
| ryansb | Eva-i: It looks to me like the behavior is correct | 20:50 |
| ryansb | both of your scenarios (one mongos & one mongod, or one mongos & two mongod's in sharded config) are not actually reliable | 20:51 |
| ryansb | because in neither case is any data replicated | 20:51 |
| Eva-i | ryansb: the program goes to "else" block if the program thinks that mongodb is reliable, right? | 20:55 |
| ryansb | it goes to the "else" if it *might* be reliable | 20:56 |
| ryansb | but in addition to the conditions of | 20:56 |
| ryansb | either a) connecting to more than one node or b) connecting via mongos | 20:56 |
| ryansb | the write concern has to be set to a durable level | 20:56 |
| *** csoukup has joined #openstack-zaqar | 20:57 | |
| Eva-i | ryansb: even if I provide a real reliable replica set to zaqar, exception will still be thrown | 20:57 |
| ryansb | oh? What type of replset did you supply? | 20:59 |
| Eva-i | ryansb: flwang1: okay, I will now create replica set on separate machines and will ask zaqar to connect to it. I will tell you the results later. | 21:01 |
| ryansb | Eva-i: k, I'm actually testing on my own box as well | 21:02 |
| ryansb | as soon as I remember how to configure a replset ;) | 21:02 |
| Eva-i | ryansb: flwang1: this replica set will consist on two mongodb servers | 21:02 |
| Eva-i | *of | 21:02 |
| ryansb | so the number of servers isn't important, you just need to have a replicaset with >1 members | 21:07 |
| *** achanda has quit IRC | 21:23 | |
| *** achanda has joined #openstack-zaqar | 21:26 | |
| *** achanda has quit IRC | 21:26 | |
| *** achanda has joined #openstack-zaqar | 21:27 | |
| ryansb | so the code seems to work properly for me in the case of using replicasets | 21:36 |
| ryansb | you may be missing the replset or writeconcern parameters in your connection URI | 21:37 |
| ryansb | uri = mongodb://kilo.local:27000,kilo.local:27001/?replicaSet=one&w=2 | 21:37 |
| ryansb | Eva-i: ^ | 21:37 |
| flwang1 | ryansb: thanks for the confirmation | 21:40 |
| ryansb | np :) | 21:41 |
| Eva-i | ryansb: 'w' parameter is write concern? | 21:42 |
| ryansb | yes | 21:42 |
| ryansb | https://docs.mongodb.org/v2.6/reference/connection-string/#connections-write-concern | 21:42 |
| *** achanda has quit IRC | 21:54 | |
| *** achanda has joined #openstack-zaqar | 21:54 | |
| Eva-i | flwang1: ryansb: I see now, shame on me. I made bug report just because I didn't knew mongodb and zaqar enough. I didn't knew that it's zaqar user who has to provide replicaSet/write concern parameters to connection via complex url. | 21:55 |
| ryansb | hey, it's okay | 21:55 |
| ryansb | now, if you want to add to our docs, or the docs on the URI option, that'd be awesome | 21:56 |
| ryansb | because if one person trips on it, it's likely that >1 person has hit it, and just floundered | 21:56 |
| ryansb | Eva-i: :) | 21:56 |
| Eva-i | ryansb: you're right, I added this as a task for docs refactoring which will happen in the near future. https://etherpad.openstack.org/p/zaqar-mitaka-docs | 22:02 |
| ryansb | Great, thank you | 22:02 |
| *** exploreshaifali has quit IRC | 22:05 | |
| *** AAzza has quit IRC | 22:05 | |
| *** AAzza has joined #openstack-zaqar | 22:05 | |
| *** AAzza has quit IRC | 22:07 | |
| *** AAzza has joined #openstack-zaqar | 22:09 | |
| *** amitgandhinz has quit IRC | 22:13 | |
| *** david-lyle has joined #openstack-zaqar | 22:16 | |
| *** boris-42 has joined #openstack-zaqar | 22:19 | |
| *** dmowrer has quit IRC | 22:22 | |
| *** ametts has quit IRC | 22:43 | |
| flwang1 | in case you haven't viewed this, two great videos about the use case of queue service and notification service https://blogs.dropbox.com/tech/2014/12/aws-reinvent-2014/ | 22:58 |
| *** csoukup has quit IRC | 23:09 | |
| *** achanda has quit IRC | 23:55 | |
| openstackgerrit | MD NADEEM proposed openstack/python-zaqarclient: Attaching a default project_id in request header for noauth authentication backend. https://review.openstack.org/243969 | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!