08:07:16 <anteaya> #startmeeting third-party
08:07:17 <openstack> Meeting started Tue Jul 28 08:07:16 2015 UTC and is due to finish in 60 minutes.  The chair is anteaya. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:07:19 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:07:21 <openstack> The meeting name has been set to 'third_party'
08:07:30 <anteaya> so tangchen do go ahead
08:07:55 <tangchen> OK
08:08:08 <eantyshev> hello!
08:08:08 <tangchen> Please take a look at here. http://paste.openstack.org/show/406017/
08:08:31 <anteaya> tangchen: why
08:08:38 <anteaya> don't just tell me to click a link
08:08:41 <anteaya> give me context
08:08:53 <tangchen> I'm now working on setting up CI system for Fujitsu. I met a proxy issue when using zuul.
08:09:01 <anteaya> thank you
08:09:08 <tangchen> I explained the context in the link.
08:09:43 <anteaya> that is less helpful in a meeting
08:09:52 <anteaya> #link http://paste.openstack.org/show/406017/
08:10:00 <anteaya> but that you for trying to be clear
08:10:10 <anteaya> so you are asking permission to post patches to zuul?
08:10:17 <anteaya> anyone can post patches to zuul
08:10:56 <tangchen> Yes, so this kind of topic should not be discussed in a meeting, right?
08:12:02 <anteaya> oh no it can be discussed
08:12:15 <anteaya> but your goal is to post patches to zuul, is it not?
08:12:25 <tangchen> Yes it is.
08:12:36 <anteaya> you don't need anyone's permission to do that
08:12:48 <anteaya> now if you had a patch to zuul and were asking for reviews
08:12:58 <anteaya> that would be something wonderful to bring to a meeting
08:13:10 <anteaya> as we can see what code you would like to introduce
08:13:17 <anteaya> and possibly review
08:13:27 <anteaya> does that make sense
08:14:47 <tangchen> OK. I can bring a patch here when I finished it. But before that, I also want some advice about how to make zuul work behind a proxy.
08:15:00 <tangchen> Has anyone met the same problem before ?
08:15:20 <kaisers> Sry, no exp with zuul / proxy
08:16:04 <anteaya> most of the people who are encountering this are in china
08:16:12 <anteaya> as china sets up a firewall
08:16:37 <anteaya> all infra tools were actually designed to not have to think about a firewall
08:16:49 <anteaya> as open source doesn't need one
08:17:03 <anteaya> and these are open source tools being used with an open source project
08:17:35 <tangchen> Oh no, I think it has nothing to do with the firewall in China. I set up the CI system on a machine in Japan.
08:17:47 <tangchen> The proxy is set by the company.
08:18:23 <tangchen> for some security reasons
08:18:33 <anteaya> oh
08:19:03 <anteaya> they really need security on throw away nodes?
08:19:18 <anteaya> all of the infra tools need their own servers
08:19:27 <anteaya> they should be on their own network
08:19:35 <anteaya> and not share server space with anything else
08:19:40 <anteaya> they aren't good neighbours
08:21:12 <anteaya> and most folks don't have zuul behind a firewall
08:21:28 <anteaya> so there will be fewer operators in the same situation
08:21:50 <tangchen> In my environment, servers cannot connect to internet directly. They should pass a proxy to limit which site can be accessed and which one cannot.
08:22:13 <eantyshev> tangchen: May I ask a small question regarding your CI setup: how do you provide logs of your CI?
08:22:49 <lennyb_> hi, sorry I am late
08:23:22 <tangchen> <anteaya>: The zuul server is in the private environment. And it has to connect to Gerrit. So I want zuul to support a proxy.
08:23:23 <anteaya> lennyb_: no problem
08:23:35 <anteaya> tangchen: I understand
08:24:02 <anteaya> tangchen: you asked if anyone else had the same issue, I'm saying it will be harder to find operators who have the same issue
08:24:23 <anteaya> tangchen: the folks from china are all behind a firewall, so might have some experience
08:24:52 <tangchen> eantyshev: I'm trying to set up a log server. But the logs may not be shared in the internet.
08:25:37 <eantyshev> tangchen: that's what I mean, you will need public server to export logs
08:26:03 <tangchen> anteaya: OK. I understood. :)
08:27:08 <anteaya> eantyshev: thanks for bringing that point up
08:27:20 <anteaya> tangchen: the point of the logs is for them to be shared on the internet
08:27:34 <anteaya> tangchen: if you don't share them, you can't comment on patches
08:27:59 <tangchen> eantyshev: Yes. If the servers are public, it is OK. But now, what I'm trying to patch is: add an interface for users to set a proxy for zuul.
08:29:58 <clarkb> you can already do tgat...
08:30:36 <clarkb> requests honors http_proxy and for ssh point zuul at a port forward or corkscrew or run with tsocks etc
08:31:27 <tangchen> I tried something similar before. It was proxychain.
08:31:38 <tangchen> But it doesn't work.
08:32:05 <anteaya> clarkb: hi, thank you
08:32:39 <tangchen> I'm not sure if it was because zuul is running background.
08:34:33 <tangchen> OK, I think it would be better to post a patch first, and then ask for review. :)
08:34:47 <tangchen> Thank you very much. :)
08:34:58 <clarkb> I dont think you need to patch anythibg is my point
08:35:10 <clarkb> proxyibg is already doable in a variety of ways
08:37:14 <tangchen> But what if I want to construct the CI system automatically ?  If I use the community version tools, and it does not support proxy, I have to maintain a tool for myself, right ?
08:38:16 <clarkb> well you dont need to maintain the tool just use it
08:38:52 <clarkb> and that way you can choose an appropriate proxy for your local situation.
08:40:48 <anteaya> tangchen: does that help?
08:40:56 <kaisers> :q
08:41:06 <kaisers> sry
08:41:11 <anteaya> tangchen: clarkb reviews zuul patches, so knows far more about zuul than I
08:41:22 <anteaya> kaisers: no problem
08:41:33 <anteaya> did anyone else have anything they wanted to discuss today?
08:41:53 <tangchen> Well, I mean the community may offer a tool to construct the CI system automatically. Just like os-ext-test. And it starts zuul without any tool, like corkscrew...
08:42:08 <tangchen> So I have to do it myself.
08:42:28 <eantyshev> Our CI occasionally reports merge failures: http://paste.openstack.org/show/406023/
08:42:32 <tangchen> If I can configure it in zuul, then it will be much easy to do.
08:42:56 <lennyb_> eantyshev: we had a discussion about it yesterday
08:43:40 <lennyb_> eantyshev: I will prepare a patch for third_party.html
08:44:23 <eantyshev> lennyb_: do you mean yesterday 3rd party meeting?
08:44:31 <lennyb_> eantyshev: http://paste.openstack.org/show/405930
08:44:50 <anteaya> eantyshev: basically akerr offered a configuration that emailed the operator on a merge failure
08:44:55 <anteaya> eantyshev: it didn't post to gerrit
08:45:07 <anteaya> tangchen: O
08:45:21 <anteaya> I'm not sure how much more help I can be to you, tangchen
08:45:35 <anteaya> I do believe that clarkb is saying you can already do what you need to do
08:45:44 <lennyb_> eantyshev: you can update akerr's sample to your needs, I am just evaluation this in our setup
08:45:55 <tangchen> anteaya: That's all right. :)
08:46:02 <tangchen> very helpful. Thanks.
08:46:11 <eantyshev> anteaya, lennyb_: I do it already, our CI won't report merge failures, but it skips some changes
08:46:55 <anteaya> tangchen: okay good, you are welcome to come back and share what you learned if you wish, it might help other operators
08:47:08 <tangchen> OK :)
08:47:08 <lennyb_> eantyshev: you said you have a merge commenting issue
08:48:32 <anteaya> eantyshev: can you expand on what you would like help with?
08:48:37 <eantyshev> lennyb_: my problem seems that our CI fails to merge something that looks like is able to be merged
08:49:04 <anteaya> have you tried manually merging the same tip, patch combination?
08:49:18 <anteaya> looks like and does can be very different
08:49:27 <eantyshev> anteaya: I disabled commenting merge failures, if it is what you mean
08:50:03 <eantyshev> anteaya: no, didn't do that manually :(
08:50:25 <anteaya> well it might not actually be able to merge
08:50:35 <anteaya> in which case it is working as expected
08:51:23 <anteaya> anything more here?
08:51:32 <kaisers> I have a q: does anybody here have exp with a manila devstack / tempest setup?
08:51:45 <anteaya> I do not
08:51:58 <anteaya> anyone else have experience with a manila setup?
08:52:09 <kaisers> ok, then i'll carry this to the manila channel afterwards.. :)
08:52:13 <anteaya> kaisers: is there a reason why you ask?
08:52:23 <kaisers> yep got issues with the network setup
08:52:33 <anteaya> have a paste?
08:52:33 <kaisers> I'm getting network errors and wondered if i missed something in the setup. Ex:  ShareNetworkNotFound: Share network wrong_id could not be found
08:52:47 <kaisers> Not sure where 'wrong_id' is coming from. :)
08:53:10 <anteaya> have you grepped the relevant repos?
08:53:16 <lennyb_> eantyshev: is it something specific? maybe you need to clean your git cache.
08:53:29 <kaisers> anteaya: not so far, will do that, thnx
08:53:34 <eantyshev> even providing our CI is triggered by "Build succeeded" comments from Openstack Jenkins, we have probability that something will be merged in between
08:53:59 <anteaya> kaisers: it is coming from manila: http://hound.openstack.org/?q=wrong_id&i=nope&files=&repos=
08:54:37 <anteaya> eantyshev: I didn't fully understand your last sentence
08:54:37 <kaisers> anteaya: ok, i'll dig into that
08:54:38 <eantyshev> anteaya: lennyb_: I will follow your suggestions, thank you! Just hoped somebody faced the same problem
08:54:47 <anteaya> eantyshev: ah okay thanks
08:54:53 <anteaya> kaisers: thanks
08:55:01 <anteaya> anything more anyone wants to discuss?
08:55:59 <anteaya> any reason why I should not end the meeting?
08:56:08 <lennyb_> eantyshev: we had a problem, when git was constantly failing with merge failure, looked like we killed zuul in the 'bad timing'. So we had to remove all git repos that zuul caches in the server
08:56:37 <lennyb_> anteaya: nothing more to discuss from me
08:56:38 <anteaya> thanks lennyb_
08:56:46 <eantyshev> lennyb_: ok, seen the same, but not this time
08:57:26 <anteaya> anything more?
08:57:56 <lennyb_> eantyshev: btw, there was some script to 'clean' git repo and removed all unused staff...maybe this will help
08:58:11 <anteaya> lennyb_: do you have a link to the script?
08:58:27 <anteaya> or is it a git command?
08:59:05 <lennyb_> anteaya: it was a tool I think, it was discussed here a month ago
08:59:30 * lennyb_ trying to look for a tool
08:59:35 <anteaya> thanks
08:59:38 * anteaya waits
09:00:51 <lennyb_> anteaya: I will try to find it offline and send to you.eantishev
09:01:16 <eantyshev> lennyb_: Thank you!
09:01:31 <anteaya> lennyb_: and perhaps link in a future meeting
09:01:41 <eantyshev> I'll wait in #openstack-infra
09:01:50 <anteaya> thanks all for your kind attendance and participation today
09:01:56 <anteaya> see you next week
09:01:59 <kaisers> thnx & bye
09:02:00 <anteaya> #endmeeting