15:00:49 <puiterwijk> #startmeeting Cockpit public meeting 2014-09-08
15:00:49 <zodbot> Meeting started Mon Sep  8 15:00:49 2014 UTC.  The chair is puiterwijk. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:49 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:50 <puiterwijk> #chair puiterwijk andreasn mvollmer stefw sgallagh
15:00:50 <zodbot> Current chairs: andreasn mvollmer puiterwijk sgallagh stefw
15:00:52 <puiterwijk> #topic Welcome
15:00:54 <puiterwijk> Okay, so let's do a roll call: who is arround here for the meeting?
15:00:58 <puiterwijk> Okay, so let's do a roll call: who is arround here for the meeting?
15:01:02 <andreasn> here
15:01:07 <sgallagh> .hellomynameis sgallagh
15:01:08 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
15:01:13 <puiterwijk> .hellomynameis puiterwijk
15:01:14 <zodbot> puiterwijk: puiterwijk 'Patrick Uiterwijk' <puiterwijk@redhat.com>
15:01:20 <andreasn> .hellomynameis andreasn
15:01:21 <zodbot> andreasn: andreasn 'Andreas Nilsson' <anilsson@redhat.com>
15:01:28 <stefw> .hellomynameis stefw
15:01:29 <zodbot> stefw: stefw 'Stef Walter' <stefw@redhat.com>
15:01:42 <puiterwijk> mvollmer: you around?
15:01:46 <mvollmer> yup
15:01:52 <puiterwijk> okay. let's start with your topic then
15:01:54 <puiterwijk> #topic Test day test cases
15:02:11 <mvollmer> .hellomynameis mvollmer
15:02:12 <zodbot> mvollmer: Sorry, but you don't exist
15:02:17 <mvollmer> :-(
15:02:21 <puiterwijk> mvollmer: you need to use your FAS username :-)
15:02:30 <mvollmer> .hellomynameis mvo
15:02:31 <zodbot> mvollmer: mvo 'Marius Vollmer' <marius.vollmer@gmail.com>
15:02:37 <mvollmer> :-)
15:02:49 <puiterwijk> great, another identity crisis solved :)
15:02:51 <mvollmer> so, test day test cases.
15:02:56 <andreasn> zodbot: at least he's a human out of flesh and blood unlike yourself
15:02:56 <zodbot> andreasn: 1410134400
15:03:12 <andreasn> ok man
15:03:33 <puiterwijk> mvollmer: yeah, we have the test day on Sept 16th, which is currently the scheduled release date for F21 Alpha
15:03:43 <stefw> which ends up being pretty dumb
15:03:45 <puiterwijk> (unless we slip one more week, that is)
15:03:48 <mvollmer> puiterwijk, is that a problem?
15:03:51 <stefw> because we can't merge the fixes that we've done recently
15:03:56 <stefw> and everyone will find the same old bugs
15:04:01 <puiterwijk> mvollmer: just a statement.
15:04:03 <stefw> can we push to updates-testing and have people pull from there for the test day?
15:04:11 <sgallagh> stefw: Yes, you can
15:04:21 <sgallagh> We can also generate a test-day install image that includes it
15:04:37 <andreasn> puiterwijk: I've heard that slipping one more week is not unlikley :/
15:04:40 <sgallagh> Furthermore, cockpit will probably get a blanket Freeze Exception to get known fixes in for the test day
15:04:40 <mvollmer> ok, let's start at the beginning: we tell people to install a image to a VM?
15:04:53 <mvollmer> sgallagh, oh, cool.
15:05:08 <puiterwijk> andreasn: that's no surprise there if you look at previous releases :)
15:05:24 <mvollmer> sgallagh, the whole F21 is like a beta test for Cockpit, kind of.
15:05:26 <stefw> mvollmer, not only vm's
15:05:48 <sgallagh> mvollmer: I think that's a bit arrogant ;-)
15:05:49 <mvollmer> stefw, ok, also on real hardware?  makes sense.
15:05:49 <stefw> if people have test hardware, we'd love for them to also do testing there
15:06:00 <stefw> sgallagh, no it's not
15:06:11 <stefw> i've been smashing my head against fedora 21 bugs for the last couple work days
15:06:30 <stefw> very few people appreciate the amount of stuff that cockpit interacts with
15:06:32 <andreasn> hopefully we'll get a lot of good data out of peoples real machines
15:06:34 <mvollmer> sgallagh, sorry, i didn't mean that the purpose of F21 is to be a beta test, but cockpit will be in beta the whole time.
15:06:34 <stefw> we end up finding everyone elses bugs too
15:06:35 <sgallagh> (sorry, I was indicating that the statement sounded like "Fedora 21 exists for Cockpit's benefit")
15:07:04 <mvollmer> sgallagh, yeah, I didn't mean that. :-)
15:07:07 <sgallagh> But yes, Cockpit is a 1.0 and F21 is its flagship deliverable :)
15:07:16 <sgallagh> s/deliverable/delivery vehicle/
15:07:27 <mvollmer> yep
15:07:53 <mvollmer> so people install F21 alpha on real hardware and are taking those risks, so we can also ask them to run Cockpit.
15:08:11 <mvollmer> (how did this zodbot info thing go?)
15:08:27 <puiterwijk> mvollmer: just prepend your message with #info
15:08:45 <mvollmer> #info People install F21 alpha on real hardware and are taking those risks, so we can also ask them to run Cockpit on real hardware.
15:09:05 <stefw> so besides basic 'poke around and do intelligent stuff with cockpit' testing ...
15:09:17 <stefw> what are the test cases we want to put together for people to run through?
15:09:28 <puiterwijk> and I guess we probably want a Cockpit-specific F21 image for test day, so we can get the latest fixes in. that should be produced a day before the test day, I think
15:09:44 <mvollmer> question: do all testers do the same things, or do we distribute test cases?
15:09:55 <stefw> well depends how many there are
15:10:05 <stefw> we may choose to have several basic tests, and then say "go wild"
15:10:07 <stefw> after that point
15:10:17 <stefw> realistically we won't have time to build more than about 10 test cases anyway
15:10:19 <mvollmer> #info we make a specific F21 with the version of Cockpit we want to test
15:10:23 <mvollmer> #undo
15:10:23 <zodbot> Removing item from minutes: INFO by mvollmer at 15:10:19 : we make a specific F21 with the version of Cockpit we want to test
15:10:32 <mvollmer> #info we make a specific F21 image with the version of Cockpit we want to test
15:10:55 <mvollmer> stefw, yes, makes sense.
15:11:00 <puiterwijk> of course there's some things we want to test with varying hardware like storage and networking I think
15:11:14 <sgallagh> Yeah, we should produce the highest-value test cases ourselves.
15:11:23 <stefw> so what are tohse?
15:11:46 * stefw thinks the best testing will come from the QE guys joining us for the day
15:11:53 <sgallagh> My top three would be 1) Storage (RAID, LVM and formatting)
15:11:58 <mvollmer> #info We prepare around ten test cases, and then ask people to go and explore.
15:12:01 <stefw> but for any others we want some basic test cases outlined, and we need to figure out what they are
15:12:08 <sgallagh> 2) Networking (basic settings, bonding)
15:12:12 <sgallagh> and 3) Container management
15:12:24 <stefw> k, that's about 20 or so test cases
15:12:33 <stefw> so i guess we can pick 2 from each of those\
15:12:41 <stefw> and then do 2 for basic login, starting cockpit
15:12:49 <stefw> or something like that
15:12:51 <andreasn> sounds good
15:13:09 <sgallagh> In the case of Fedora Server, it should be available by default.
15:13:22 <puiterwijk> yeah, it is.
15:13:31 <mvollmer> #info two each of Basic (login, logout), Storage, Network, Docker.
15:13:35 <sgallagh> Sorry, that was unclear. I was noting that as a test case
15:13:41 <puiterwijk> sgallagh: ah, right
15:13:51 <puiterwijk> #info Started by default should be tested in F21
15:13:58 <sgallagh> #undo
15:13:58 <zodbot> Removing item from minutes: INFO by puiterwijk at 15:13:51 : Started by default should be tested in F21
15:14:08 <sgallagh> #info Started by default should be tested in F21 Server
15:14:43 <mvollmer> how interactive is the test day going to be?
15:14:48 <sgallagh> (As an aside, the test day image should be the Fedora Server image plus latest Cockpit or something else?)
15:14:59 <stefw> interactive?
15:14:59 <sgallagh> mvollmer: Depends on the level of participation
15:15:10 <mvollmer> right
15:15:18 <sgallagh> The project maintainers are expected to be available to answer questions
15:15:21 <stefw> it's usually about lots of people on IRC and helping them file issues, etc.
15:15:25 <mvollmer> stefw, I meant, do people show up and tell us in real-time what they are doing?
15:15:32 <stefw> often they do
15:15:37 <mvollmer> right, nice
15:15:41 <stefw> we build a table where they post their results for test cases too
15:15:48 <stefw> so some people just file stuff there, without making much noise
15:15:59 <sgallagh> stefw: There's a new tool for that which is nicer than the old Wiki-edit approach
15:16:41 <stefw> are you gonna tell us what it is?
15:16:58 <sgallagh> Sorry, I was trying to find the link
15:17:04 <stefw> ah, sorry
15:17:18 <sgallagh> No problem. I'm hitting a wall. I'll check with QA where it's kepy
15:17:21 <sgallagh> *kept
15:17:56 <puiterwijk> sgallagh: you're not talking taskotron, right? as that's automated testing
15:18:11 <sgallagh> http://testdays.qa.fedoraproject.org/testdays/all_events
15:18:25 <puiterwijk> ah, right.
15:18:33 <sgallagh> Recent example: http://testdays.qa.fedoraproject.org/testdays/show_event?event_id=17
15:18:58 <stefw> puiterwijk, can you help us get listed there?
15:19:26 <puiterwijk> stefw: I think it's generated from the wiki, but yeah, I can look into that
15:19:38 <sgallagh> puiterwijk: It's not automatically generated, I'm pretty sure.
15:19:49 <puiterwijk> sgallagh: it's from metadata in the wiki page, IIRC
15:19:53 <sgallagh> ok
15:20:25 <puiterwijk> anyway, I can look into that
15:20:35 <mvollmer> ok, I think I can write some test cases now.
15:20:57 <mvollmer> #action mvollmer write tenish test cases as outlined above
15:21:00 <puiterwijk> mvollmer: okay, maybe it'd be an idea if you'd send a mail to the list once you have something to solicit review?
15:21:07 <mvollmer> sure
15:21:18 <puiterwijk> #action puiterwijk Get us listed in the testdays.qa.fedoraproject.org/testdays app
15:21:24 <stefw> mvollmer, let me know if you need help ... i just wish the f21 pain would stop
15:21:37 <mvollmer> stefw, I think it's quite ok, actually.
15:21:51 <puiterwijk> We also need someone that can generate an image on Sept 15th
15:21:57 <stefw> well, i can't even get our integration test suite to run at the moment
15:22:05 <stefw> each day brings a new set of bugs :S
15:22:33 <mvollmer> stefw, you mean running VERIFY inside a F21 install?  Or running with TEST_OS=fedora-21
15:22:39 <stefw> inside f21
15:22:40 <sgallagh> stefw: Are you running it against the stable or updates-testing-enabled version of F21?
15:22:41 <stefw> used to work fine
15:22:49 <mvollmer> stefw, right, never tried that. :-)
15:23:06 <stefw> updates testing is enabled
15:23:07 <sgallagh> Now that we're in freeze, I'd recommend 'distro-syncing' to the stable packages
15:23:09 <stefw> i guess i can try reverting
15:23:15 <sgallagh> It'll reduce the churn
15:23:17 <stefw> good idea
15:24:08 <mvollmer> ok, next topic?
15:24:19 <puiterwijk> well, we still need to decide who will generate the image next week?
15:24:32 <mvollmer> right
15:24:35 <puiterwijk> sgallagh: do you have experience with that?
15:24:53 <stefw> i think it changes from release to release
15:24:59 <mvollmer> worst case: we use the F21 alpha image and tell people to "yum update cockpit"
15:25:00 <stefw> experience might not be that useful
15:25:01 <puiterwijk> stefw: well, I found the SOP..
15:25:19 <puiterwijk> at least, somewhere
15:25:24 <puiterwijk> ah, yeah:
15:25:25 <stefw> i've done it before, using lorax ... but it took me a long long time to make it work
15:25:26 <puiterwijk> #link https://fedoraproject.org/wiki/QA/Test_Days/Live_Image
15:25:40 <puiterwijk> stefw: this is an easy SOP how to create the image
15:25:58 <puiterwijk> although this is for a live image
15:26:00 <stefw> yup, that's the one i did
15:26:27 <puiterwijk> Do we want a live disc, or an installable disc?
15:26:35 <puiterwijk> I think live disc is a lot easier to test for people
15:26:38 <sgallagh> I can easily regenerate the Server install DVD
15:26:46 <sgallagh> Takes about 90 minutes
15:26:59 <sgallagh> It's rather large, though. 1.9GB
15:27:15 <sgallagh> I don't know if we want to look into creating a smaller package set just for the Test Day
15:27:19 <puiterwijk> sgallagh: okay, so could you generate one next week arround the 15th? at least so it's available on the 16th?
15:27:34 <sgallagh> If we do, please get me a list of our important deps and I can try some test composes this week
15:28:28 <puiterwijk> #action sgallagh to spin a F21 Server install disc for the test day on September 16th
15:28:53 <puiterwijk> sgallagh: well, I think most of it is in our Requires (implicit and explicit). I'd say if we miss something there, it's a bug?
15:28:56 <sgallagh> Do we want the full disk then?
15:29:03 <stefw> puiterwijk, i don't think that's the case
15:29:05 <sgallagh> puiterwijk: We don't Requires: docker-io
15:29:07 <sgallagh> (For example)
15:29:12 <stefw> right, exactly
15:29:28 <puiterwijk> hum, right. so we need a list of those packages
15:29:28 <andreasn> as close as possible to the default server product install would be best
15:29:31 <sgallagh> Cockpit seems to use whatever is installed, but not force it
15:29:45 <sgallagh> andreasn: Ok, that's easy then. I just respin the Server DVD
15:30:03 <puiterwijk> sgallagh: okay, thanks. do you have enough quota on fpeople to put it there, or do you need me to bump your quota?
15:30:18 <puiterwijk> (or do you want me to create another project where you can put it?)
15:30:19 <andreasn> sgallagh: sounds good
15:31:06 <sgallagh> puiterwijk: I think I can get hosting from QA for the test day DVD
15:31:29 <sgallagh> I'll check on that
15:31:32 <puiterwijk> sgallagh: okay. let me know if you don't, and we'll have other places where we can put it
15:32:14 <puiterwijk> Anything else on the test days?
15:32:37 <puiterwijk> #topic Fedora 21 Status
15:32:41 <puiterwijk> Most of this we already discusses
15:32:44 <puiterwijk> but for reference:
15:32:45 <puiterwijk> #info Fedora 21 Alpha is no-go for now -> One week slip
15:32:47 <puiterwijk> #info Current scheduled Alpha release: September 16th
15:33:13 <andreasn> Fedora 21 Beta milestone is at 87%
15:33:27 <puiterwijk> great!
15:33:49 <mvollmer> heh, let's not start the middle management number fetish...
15:34:14 <puiterwijk> mvollmer: nope, but sometimes random numbers are fun to mention :)
15:34:19 <stefw> mvollmer, ++
15:34:24 <mvollmer> (i can move some old closed issues to the milestone...)
15:34:40 <andreasn> indeed
15:34:52 <andreasn> I just wanted to say that things are in good shape as far as I can see
15:34:59 <puiterwijk> okay, so do we have any things that should be considered F21 blockers?
15:35:51 <andreasn> most things on the list are nice-to-have-polish
15:36:32 <andreasn> I would only block on things like if Cockpit don't start by default or something
15:36:35 <stefw> i think running the test suite would be a blocker ... mvollmer has done some work there, and i'm working on that too.
15:38:41 <puiterwijk> stefw: fair enough. do you have bugs for the things in bugzilla?
15:38:49 <puiterwijk> (the bugs you found so far)
15:38:52 <mvollmer> stefw, so, for example, DOS extended partitions must be fixed?
15:39:15 <mvollmer> puiterwijk, yes, we found three.
15:39:36 <puiterwijk> mvollmer: okay, and are they marked as proposed Blocker?
15:39:44 <mvollmer> no.
15:39:59 <puiterwijk> well, if we want to have F21 release block on it, they should be marked as such
15:40:19 <stefw> i'm not sure we need to have blockers for all the bugs we've found in f21 (whether via the integration tests or not)
15:40:40 <stefw> but just things that keep us from running the test suite on fedora 21 are serious
15:40:57 <puiterwijk> right, but you should at least request it for those that block the testsuite
15:41:05 <sgallagh> puiterwijk: Only if we want Alpha to block on it
15:41:15 <stefw> and i don't think we want to block alpha anyway
15:41:16 <sgallagh> puiterwijk: If it can wait until Beta, don't bother.
15:41:19 <stefw> so we have time
15:41:24 <puiterwijk> sgallagh: I think the BetaBlockerProposed tracker has also been created already, not?
15:41:29 <sgallagh> The tree will open back up once we have an Alpha Compose
15:41:42 <sgallagh> puiterwijk: It's there, but not being processed
15:41:44 <puiterwijk> right, but we want the people maintaining those packages to fix them, not?
15:41:59 <puiterwijk> sgallagh: right, I just meant for keeping track of it
15:42:12 <sgallagh> Yes, file the bugs and propose them for appropriate blocker/FE requests, please
15:42:33 <puiterwijk> yes, that's what I meant. then the maintainers know we want it fixed, and know what timeline
15:43:26 <mvollmer> as far as blockers in cockpit go: the non-root / wheel situation.
15:43:42 <stefw> mvollmer, yeah, i agree
15:43:51 <mvollmer> also: hostname validation rules.
15:43:57 <puiterwijk> mvollmer: is that an alpha blocker? or beta/final?
15:44:02 <mvollmer> beta/final
15:44:32 <andreasn> is there an issue open regarding the non-root case?
15:44:41 <mvollmer> sgallagh, puiterwijk, we should file bugs against cockpit in bugzilla.redhat.com for these, right?
15:44:47 <puiterwijk> mvollmer: yes.
15:44:51 <mvollmer> right
15:44:59 <sgallagh> mvollmer: Or against the packages Cockpit depends on, if the bug is there
15:45:02 <andreasn> is it this one? https://github.com/cockpit-project/cockpit/issues/915
15:45:05 <mvollmer> of course.
15:45:22 <puiterwijk> mvollmer: and also please request the BetaBlocker/FinalBlocker so we/QA/other maintainers know we want to get it in by then
15:45:36 <puiterwijk> (well, if you think we should block beta or final on that, that is)
15:45:54 <mvollmer> ok.
15:46:42 <mvollmer> #action mvollmer file blocker bugs in Cockpit on Fedora bugzilla.  (non-root / wheel situation, others)
15:46:55 <puiterwijk> any further F21 Cockpit status stuff at this moment?
15:47:10 <mvollmer> andreasn, yes, basically, things don't work so well as non-root.
15:47:19 <sgallagh> Just the usual cheerleading from me that people are excited about it.
15:47:28 <andreasn> :)
15:47:38 <puiterwijk> sgallagh: we all are I'd say :)
15:48:05 <puiterwijk> Okay, so next on my list
15:48:10 <puiterwijk> #topic Atomic feature review
15:48:13 <mvollmer> we had a high level guy in the office talking in the next room, and there was even mention of cockpit...
15:48:28 <stefw> this is a blocker for non-root: http://lists.freedesktop.org/archives/dbus/2014-September/016314.html
15:48:29 <mvollmer> "cockpit is nice, but"
15:48:32 <puiterwijk> #undo
15:48:32 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x10962e50>
15:48:43 <stefw> so ... we're pretty far away from it working completely
15:48:52 <andreasn> mvollmer: what was the but?
15:48:56 <puiterwijk> mvollmer: cool! I've been asked to give a presentation on Cockpit in the AMS office
15:49:13 <mvollmer> andreasn, didn't listen carefully enough! :-)
15:49:33 <andreasn> fair enough
15:49:46 <mvollmer> stefw, yeah, the alternative to making it work is to be clear that it doesn't work.
15:49:53 <stefw> exactly
15:49:53 <mvollmer> remove all the roles stuff, etc.
15:50:18 <puiterwijk> mvollmer: well, at least hide it from the UI?
15:50:27 <mvollmer> puiterwijk, yep.
15:50:29 * sgallagh has to run
15:50:33 <mvollmer> stefw, does "sudo" work?
15:50:41 <stefw> mvollmer, pkexec does
15:50:47 <puiterwijk> sgallagh: okay, thanks for attending. minutes will be in the list
15:51:01 <stefw> mvollmer, we should talk about the non-root situation more
15:51:05 <mvollmer> right, so we can escalate to root, but not according to sudo policies?
15:51:16 <mvollmer> stefw, ok.
15:51:17 <stefw> because part of it doesn't really ever get solved
15:51:37 <stefw> if you can log in as non-wheel for example, or with custom polkit policies, you might not have access do no X regardless of being wheel or not
15:51:42 <stefw> and we need a good story for that in cockpit
15:51:46 <stefw> maybe not all implemented now
15:51:57 <stefw> but somehow we need to represent that, make user aware of what's going on ...
15:52:08 <stefw> so the solution doesn't necessarily need to be interim
15:52:13 <puiterwijk> stefw: do you want to discuss it now, or put that on the agenda for next week?
15:52:20 <stefw> we can discuss it elsewhere
15:52:35 <puiterwijk> okay, maybe send it to the list and put it on next week's agenda to review?
15:52:45 <stefw> sure
15:52:48 <puiterwijk> ("the list" being mailing list)
15:53:16 <puiterwijk> #action stefw to send email to the list regardining non-root/wheel situation
15:53:25 <puiterwijk> #action puiterwijk to put non-root/wheel on the agenda next week
15:54:01 <mvollmer> #action mvollmer finally figure out what a hostname is and fix our dialog accordingly
15:55:12 <puiterwijk> okay, any more F21 stuff?
15:55:26 <puiterwijk> #topic Atomic feature review
15:55:32 <puiterwijk> andreasn: your point :)
15:55:35 <puiterwijk> (again
15:55:36 <puiterwijk> )
15:55:44 <andreasn> yeah, so stefw gave some feedback on the feature pages
15:55:50 <andreasn> and so gave julim, going through that
15:56:03 <andreasn> also did some investigation on general docker storage and stuff
15:56:33 <puiterwijk> #info review from stefw and julim on Atomic feature pages came in, andreasn going through
15:56:37 <andreasn> but more feedback would be awesome. So if mvollmer and puiterwijk could read through the pages and make sure it sounds sane
15:57:11 <puiterwijk> #action mvollmer puiterwijk to read through atomic feature pages and tell andreasn if it's sane
15:57:13 <mvollmer> i read them very quickyl, but I'll do that once more.
15:57:19 <andreasn> thanks!
15:57:52 <andreasn> yep, that's all for that
15:57:55 <puiterwijk> okay
15:57:58 <puiterwijk> #topic Open Floor
15:58:19 <puiterwijk> Okay, I wanted to throw one thing up: andreasn last week you said the new navigation was under review. anything from that yet?
15:58:51 <andreasn> yeah, I'm going to meet with julim, stefw and LHinson1 about that tomorrow afternoon. If anyone else wants to join, feel free
15:59:05 <puiterwijk> okay, wherei s the meeting?
15:59:12 <puiterwijk> channel/time?
15:59:16 <andreasn> conf video call
15:59:28 <puiterwijk> andreasn: hmm, not sure if that's really intended for the public
15:59:36 <puiterwijk> how about we just say people interested in joining contact you?
15:59:39 <andreasn> sure
15:59:54 <puiterwijk> #info people interested in attending conference on new navigation should contact andreasn
16:00:06 <puiterwijk> anyone else got any last points?
16:00:13 <stefw> why is files.cockpit-project.org so very very slow?
16:00:25 <puiterwijk> stefw: hmm, you mean network or CPU?
16:00:36 <stefw> sorry, i mean download speed
16:00:43 * stefw does a test
16:00:44 <puiterwijk> right, that sounds like network.
16:01:08 <puiterwijk> since the CPU and mem are not an issue at all: the server is basically sleeping
16:01:29 <puiterwijk> but yeah, I have to say it feels a bit sluggish at times. I'll ask the networking team why that is
16:01:47 <mvollmer> ohh, and I was blaming the office network...
16:01:49 <puiterwijk> but if you can quantify, that would be great?
16:02:05 <stefw> how do we quantify this? multiple download points?
16:02:24 <puiterwijk> stefw: average download speed of a big file?
16:02:37 <puiterwijk> mvollmer: same here
16:03:07 <puiterwijk> I'll run tests from some more servers and see if it's also so slow from the internal nets
16:03:50 <stefw> very odd behavior
16:04:01 <puiterwijk> stefw: is it consistently slow, or changing?
16:04:03 <stefw> if i download directly, this file takes 7 hours to download:
16:04:03 <stefw> http://files.cockpit-project.org/testdata/images/cockpit-fedora-20-x86_64-root.xz
16:04:07 <stefw> pretty consistent
16:04:15 <mvollmer> ouch
16:04:21 <mvollmer> couple minutes here.
16:04:22 <puiterwijk> hum, okay, that sounds not okay. let me ask net guys
16:04:22 <stefw> whereas if i first download to a machine in a data center in the hague
16:04:29 <stefw> then it's just 17 minutes
16:04:45 <stefw> obviously that's not fast in any case, but very different speeds
16:04:52 <puiterwijk> stefw: are you connected to the VPN?
16:04:54 <stefw> yes
16:05:03 <puiterwijk> could you try it when disconnected?
16:05:10 <stefw> sure
16:05:14 <puiterwijk> I'm starting to feel that might have an impact because of routing
16:05:38 <mvollmer> 0  466M    0 1603k    0     0  62705      0  2:09:54  0:00:26  2:09:28 37821
16:05:41 <mvollmer> oho.
16:05:47 <puiterwijk> I'm downloading at about 500KB/s
16:05:48 <mvollmer> sorry, no headers.
16:05:53 <puiterwijk> (eta 22m)
16:06:01 <mvollmer> speed 37k, est time 2:09
16:06:05 <mvollmer> hours
16:06:07 <stefw> no vpn, 9 hours
16:06:13 <puiterwijk> mvollmer: wow, 37k is way too low. are you on the VPN?
16:06:20 <mvollmer> yes
16:06:27 * stefw gets a mtr trace
16:06:56 <stefw> downloaded to a data center in 3 minutes 44 seconds in the hague
16:07:12 <mvollmer> okay, much better now:  0:05:39 1522k
16:07:22 <puiterwijk> mvollmer: that's after disconnecting from VPN? or just a retry?/
16:07:35 <mvollmer> just a retry
16:08:05 <puiterwijk> hmm, weird. okay, I'll go into it, and ask you for more checks soon
16:08:18 <puiterwijk> #action puiterwijk to look at slowness of files.cockpit-project.org network paths at times
16:08:31 <mvollmer> thanks.
16:08:43 <mvollmer> ok, signing out...
16:08:48 <mvollmer> good meeting, thanks!
16:08:51 <andreasn> thanks!
16:08:53 <puiterwijk> okay, thanks to everyone for the meeting!
16:08:55 <stefw> puiterwijk, i get lots of dropped packets to files.cockpit-project.org
16:09:10 <stefw> about 15%
16:09:11 <puiterwijk> stefw: hmm, could you fpaste me a traceroute?
16:10:01 <stefw> done
16:10:17 <puiterwijk> got it, thanks.
16:10:52 <puiterwijk> so the loss seems to happen at the PHX2 gateway, I'll ping our net guys with this info
16:11:17 <puiterwijk> any other remarks for the open floor?
16:11:49 <andreasn> nope
16:11:58 <puiterwijk> I'm closing the meeting in a minute if nothign else comes up
16:12:40 <andreasn> heading out for a walk. Later!
16:12:44 <puiterwijk> okay, thanks all for being here for this meeting! Minutes will be sent to the list in an hour approximately!
16:12:47 <puiterwijk> #endmeeting