IRC logs for #baserock for Thursday, 2015-03-26

persiastraycat: Thanks for bringing my attention to that.  I'll try to find some time to make a fuss, but it will take me a few days to do so.01:04
persiaNote that we can create a 0.x branch at any time later, if we need one for some reason.01:05
persia(as long as we know from which commit we are now working on 1.x)01:05
*** zoli__ has joined #baserock05:35
*** zoli__ has quit IRC06:23
*** zoli__ has joined #baserock06:23
*** zoli__ has quit IRC06:28
* straycat nods07:35
*** sherm_ has joined #baserock07:51
Kinnisonstraycat: yeah (re: reviewing a series) I asked and it was suggested such review comments should go on the tip commit as it is what others have been doing08:02
straycatoh, my session's expired, oops08:03
KinnisonYeah that angered me this morning too08:03
KinnisonWe need to ask ssam2 if there's any chance he can increase the expiry time08:04
straycatthe thing that annoys me most is that it shows the change metadata rather than the change itself08:04
straycatthis is what makes reviewing a series in a topic difficult08:04
straycatit's entertaining given all the empty space on my screen08:05
straycatwell it'd be entertaining if i didn't have to actually use it08:05
Kinnisonsadly the internets suggests ticking a 'remember me' box on sign-in which I cannot find08:10
*** gfinney has joined #baserock08:13
*** a1exhughe5 has joined #baserock08:18
*** mdizzle has joined #baserock08:43
*** jonathanmaw has joined #baserock08:45
*** mdunford has joined #baserock08:54
*** bashrc has joined #baserock09:02
*** rdale has joined #baserock09:13
*** tiagogomes_ has joined #baserock09:25
*** mdunford has quit IRC09:30
pedroalvarezok, if a patch has been merged, but not using gerrit, and the patch is also is in gerrit. should we abandon it before someone merges it?09:35
pedroalvarezI guess the answer is "yes"09:35
straycatyes, i don't have any powers to merge in gerrit09:35
straycator do anything, either that or i couldn't find out how to abandon the change09:36
straycatso i just merged it in git09:36
tlsahow do you debug a chunk build that morph thinks worked, but it didn't work because the software doesn't end up in the system?09:37
pedroalvarezstraycat: do you want to be in the mergers group? You are a clear candidate09:37
* jjardon thinks we should use gerrit exclusively to avoid future confusions09:38
* straycat thinks not09:38
pedroalvareztlsa: I normally put " - exit 1"  at the bottom09:38
pedroalvareztlsa: so it fails building, and you can debug the staging area09:38
pedroalvareztlsa: I mean, `exit 1` after all the install commands in the chunk morphology09:39
tlsapedroalvarez: so you need to add a chunk morph?09:39
tlsathis one doesn't have one at the moment09:40
pedroalvarezoh, then yes09:40
De|ta'wireless conneciton light'09:41
De|taah, wrong window. sorry09:41
tlsapedroalvarez: thanks09:41
tlsawhere can I look up the "default" build/install commnds09:42
*** edcragg has joined #baserock09:42
straycatpedroalvarez, i don't know yet, i'm really against moving to gerrit, primarily on the basis that it clearly doesn't support baserock workflow, but also on the basis that it has no decent tooling09:42
jjardonstraycat: you merged the stage1 branch but have the comments of the review (in gerrit) been addressed?09:43
straycati don't know09:43
straycatit has > 2 +1s, so it's been merged09:43
bashrcadmittedly the tooling is not the greatest09:44
bashrcbut it may "just" be a matter of reading the API and getting around to writing the relevant elisp09:45
* straycat wants to try gertty once we get https09:48
straycatif that turns out to be even half good then it's probably quite easy for us to help them improve it09:49
straycatbut right now there's no obviously good tooling and yet the project's already putting all its patches in gerrit, for no obviously good reason, there's no premerge testing happening currently as far as i'm aware, basically, i think we're rushing into this without thinking about it09:50
straycatwe don't even have email notifications coming in09:52
* SotK just wishes Gerrit would stop signing him out every morning09:55
bashrcthat's a point. There are no notifications as such09:56
KinnisonSotK: aye, increasing that expiry thing would be handy09:57
Kinnisonbashrc: email would be possible if we had an MTA available in Baserock09:57
* Kinnison was going to suggest msmtp to ssam2 when he next turns up09:58
bashrcso presumably the process is that after a specifiable number of +1's whoever is the maintainer of each gerrit repo will merge to master09:58
jjardonI agree there are still some missing functionality, and the only thing i can think to solve them is to offer our help to the infrastructure team as much as we can (and that can be as simple as file bugs in our storyboard, instead complaining here)09:59
bashrcbaserock doesn't have a MTA?10:00
bashrcthere is a mailing list10:00
pedroalvarezjjardon: couldn't agree more10:00
rjekExim 4.85 with the same build feature set as Debian's exim-daemon-heavy, please :)10:01
* rjek hides10:01
pedroalvarezSam was integrating exim, but didn't have time to finish it10:01
bashrcwhile a MTA may not be needed for many embedded systems it will be needed for most server systems10:02
rjekThe "heavy" feature set has a /lot/ of dependancies10:02
rjekSpamAssassin and ClamAV, for example, as well as SPF and DKIM libraries.10:03
bashrcexim is hideously complicated, but such is the way of MTAs10:03
rjekIt's basically a Malbolge interpreter with implementations of SMTP.10:04
jonathanmawhrm, I'm trying to build icu in qt5-tools-jetson, and the build fails in some rather strange ways. From morph, the build fails at "genrb: error while loading shared libraries: lib/ internal error" when I chroot into the failed build root, 'make' segfaults, instead.10:04
jonathanmawinterestingly, running the 'make' inside the failed dir from outside a chroot works fine.10:05
pedroalvarezjonathanmaw: oh.. make segfaulting...10:05
jonathanmawwhich is weird, since it managed to build everything else up to that point.10:05
richard_mawthat's rather odd, I'd first try wiping out your hardlink cache10:06
bashrcrjek: turning the configuration system into its own language seems like a fail to me #turingcomplete #securitynightmare10:06
straycatjjardon, it is not nearly that simple, we don't know what gertty is like and if it's awful we're basically going to have to write our own tools10:06
rdalemake seg faulting in a chroot is because you need to bind mount /dev /proc and /tmp i think10:06
pedroalvarezjonathanmaw: make segfaulting in the staging area is a known error, though10:06
rjekbashrc: Only the adminstrator has access to the turing-complete parts.10:06
rjekbashrc: The administrator, should they wish, can run arbitrary C code too.10:07
rjekSo no, it's not a security nightmare.10:07
jonathanmawrichard_maw: that's /src/tmp/chunks, right?10:07
richard_mawjonathanmaw: aye10:07
*** ssam2 has joined #baserock10:11
*** ChanServ sets mode: +v ssam210:11
radiofreejonathanmaw: i merged a fix for that yesterday10:11
radiofreewell, ssam2/gerrit did10:12
straycatssam2, can you please consider the backlog10:16
straycatthis project seems to be ignoring me now10:16
ssam2what backlog?10:17
ssam2i'm 3 weeks behind with review10:17
ssam2if that's the backlog you mean10:17
straycatno, the irc log10:17
ssam2nope, please send email if it's important10:18
ssam2irc is transient as far as I'm concerned10:18
straycatoh ffs10:18
jjardonstraycat: really i think that's a personal opinion, tools are good enough for some of us already10:18
straycatthis is not personal10:18
straycati've already stated that i cannot deny the importance of premerge testing10:19
straycati'm asking that we get a chance to try out the tooling we've not yet been able to try before we try to switch10:19
straycati get that some of you are happy with a switch to a web interface10:20
straycatsome of us aren't10:20
straycatand we're being alienated10:20
ssam2if this about gerrit, there has been a test gerrit instance set up for months10:20
ssam2gerrit has been being discussed for months10:20
richard_mawlet's not argue about this now10:20
straycatssam2, Not everyone on this project has the LUXURY of being paid to focus on those things10:20
richard_mawI'll continue to do mailing list review until we no longer get anything in by that channel.10:20
straycatmost of my focus on this sort of thing is at the weekend10:20
straycatthat time is extremely limited10:20
radiofreeis mason ok? it went red at 01:37:3810:21
radiofreealso maybe it would be useful to add to the topic?10:23
pedroalvarezradiofree: no, we have 2, and we want to have more. The topic would be too long10:23
pedroalvarezstraycat: we are going to get the chance to test gerrit with all its features before we vote10:24
pedroalvarezat least is what i understood fromprevious discussions10:24 :)10:25
KinnisonAnyone here have a stage2-gcc artifact set I could take a squiz at?10:39
* Kinnison would settle for a file listing from each of the artifacts10:39
*** paulwaters_ has joined #baserock10:41
*** paulwaters_ has joined #baserock10:41
paulsherwoodKinnison: what about
KinnisonI'd need to know artifact IDs10:44
* Kinnison just wants to look at the file listings10:45
ssam2`morrh list-artifacts REPO REF system` can tell you the artifact IDs10:46
ssam2'morph' rather10:46
Zarahm, getting this error, not sure why: ERROR: Git directory /src/cache/gits/git___git_baserock_org_baserock_baserock_definitions has no commit at ref 8932bad72a43be76473c9c35accf1a7d3ef0344e^{commit}10:49
Zaraoh wait, mason10:49
*** paulwaters_ has quit IRC10:49
Zarathat's unfortunate, I thought it was an issue with the stuff on my baserock system and deleted the stuff in the cache ;_;10:49
bashrcok so cliapp finds the firehose path, but doesn't find the plugin file10:57
Zara(do we know what's up with definitions at the moment?)10:59
SotKwhat do you mean?11:00
ZaraI got the error I posted above, checked mason and it says it's down11:00
paulsherwoodmason fail, iexpect11:00
* paulsherwood fears it'll be his patch :/11:01
paulsherwoodurghh. it is11:01
pedroalvarezpaulsherwood: I don't think so11:01
paulsherwoodooh, interesting! :)11:02
pedroalvarezmason x86_64 i still building11:02
Zaraah, okay11:02
ZaraIf that's not the cause of my error then I don't know what's going wrong. :S11:03
SotKZara: what command are you running to get that error?11:03
jonathanmaw:-/ Rebased against master to get radiofree's fix to libicu, and I'm rebuilding from libstdc++11:13
*** simonh__ has joined #baserock11:26
*** a1exhughe5 has quit IRC11:30
*** simonh_ has quit IRC11:30
*** perryl has quit IRC11:30
*** pdar has quit IRC11:30
*** edcragg has quit IRC11:30
*** mdizzle has quit IRC11:30
*** gfinney has quit IRC11:30
*** robtaylor has quit IRC11:30
*** bjdooks has quit IRC11:30
*** rdale has quit IRC11:30
*** radiofree has quit IRC11:30
*** ratmice__ has quit IRC11:30
*** ssam2 has quit IRC11:30
*** jmacs has quit IRC11:30
*** persia_ has quit IRC11:30
*** tiagogomes_ has quit IRC11:30
*** CTtpollard has quit IRC11:30
*** flatmush has quit IRC11:30
*** pedroalvarez has quit IRC11:30
*** ChanServ has quit IRC11:30
*** sherm_ has quit IRC11:30
*** DavePage has quit IRC11:30
*** richard_maw has quit IRC11:30
*** dabukalam has quit IRC11:30
*** cyndis has quit IRC11:30
*** petefoth has quit IRC11:30
*** wdutch has quit IRC11:30
*** benbrown_ has quit IRC11:30
*** De|ta has quit IRC11:30
*** jjardon has quit IRC11:30
*** straycat has quit IRC11:30
*** juergbi has quit IRC11:30
*** bashrc has quit IRC11:30
*** jonathanmaw has quit IRC11:30
*** kejiahu has quit IRC11:30
*** SotK has quit IRC11:30
*** sambishop has quit IRC11:30
*** fay_ has quit IRC11:30
*** paulw has quit IRC11:30
*** persia has quit IRC11:30
*** Krin has quit IRC11:30
*** inara` has quit IRC11:30
*** Zara has quit IRC11:30
*** mwilliams_ct has quit IRC11:30
*** bwh has quit IRC11:30
*** petefotheringham has quit IRC11:30
*** rjek has quit IRC11:30
*** paulsherwood has quit IRC11:30
*** ctgriffiths has quit IRC11:30
*** JPohlmann has quit IRC11:30
*** a1exhughe5 has joined #baserock11:34
*** bjdooks has joined #baserock11:34
*** perryl_ has joined #baserock11:34
*** mdunford has joined #baserock11:34
*** zoli__ has joined #baserock11:34
*** ssam2 has joined #baserock11:34
*** edcragg has joined #baserock11:34
*** tiagogomes_ has joined #baserock11:34
*** rdale has joined #baserock11:34
*** bashrc has joined #baserock11:34
*** jonathanmaw has joined #baserock11:34
*** mdizzle has joined #baserock11:34
*** gfinney has joined #baserock11:34
*** sherm_ has joined #baserock11:34
*** cyndis has joined #baserock11:34
*** pdar has joined #baserock11:34
*** Krin has joined #baserock11:34
*** robtaylor has joined #baserock11:34
*** wdutch has joined #baserock11:34
*** petefoth has joined #baserock11:34
*** fay_ has joined #baserock11:34
*** radiofree has joined #baserock11:34
*** jmacs has joined #baserock11:34
*** paulw has joined #baserock11:34
*** CTtpollard has joined #baserock11:34
*** inara` has joined #baserock11:34
*** flatmush has joined #baserock11:34
*** kejiahu has joined #baserock11:34
*** Zara has joined #baserock11:34
*** benbrown_ has joined #baserock11:34
*** mwilliams_ct has joined #baserock11:34
*** De|ta has joined #baserock11:34
*** paulsherwood has joined #baserock11:34
*** SotK has joined #baserock11:34
*** pedroalvarez has joined #baserock11:34
*** DavePage has joined #baserock11:34
*** petefotheringham has joined #baserock11:34
*** bwh has joined #baserock11:34
*** persia has joined #baserock11:34
*** sambishop has joined #baserock11:34
*** persia_ has joined #baserock11:34
*** jjardon has joined #baserock11:34
*** ratmice__ has joined #baserock11:34
*** straycat has joined #baserock11:34
*** richard_maw has joined #baserock11:34
*** dabukalam has joined #baserock11:34
*** sets mode: +vv ssam2 pedroalvarez11:34
*** ctgriffiths has joined #baserock11:34
*** juergbi has joined #baserock11:34
*** JPohlmann has joined #baserock11:34
*** ChanServ has joined #baserock11:34
*** rjek has joined #baserock11:34
*** sets mode: +o ChanServ11:34
ZaraSotK: yeah, I tried removing it, didn't work. seems it's trying to build from master rather than from my local branch11:34
Zaraand master doesn't have my local commit on it, as you'd expect :P11:34
bashrcrichard_maw: the firehose plugin is detected but not parsed by cliapp11:34
Zaraso it's not building11:34
richard_mawbashrc: what do you mean by "not parsed by cliapp"?11:34
bashrcrichard_maw: it looks like there may be bugs in python inspect.getmembers11:34
Zaraso, how do I get it to try to build from my local branch?11:34
radiofreepedroalvarez: why can't they go on the same network as cache?11:35
bashrcrichard_maw: cliapp uses load_plugin_file to extract class information. It fails on firehose11:35
radiofreepedroalvarez: this sounds like a job for baserock ops?11:35
* radiofree isn't sure who they are, of if they exist11:36
richard_mawbashrc: do you have a link to the code of the firehose plugin?11:36
pedroalvarezwell, cache is running in a openstack tenant, and I believe that we can't plug the jetsons there11:36
radiofreecan't we just do what we did last time?11:36
pedroalvarezradiofree: last time they had publics Ip, but I believe that we can't do that this time11:37
radiofreeand disable password authentication for ssh11:37
bashrccliapp seems to get confused when parsing that11:38
pedroalvarezradiofree: I suggested public IPs to them, but the answers was like a "no"11:38
radiofreewho did you suggest it to?11:38
pedroalvarezradiofree: people of DC11:39
radiofreeany particular reason?11:39
radiofreei'm assuming the moonshot server will have a public ip at some point, what's the difference between that and a jetson, the form factor?11:39
bashrcrichard_maw: also see load_plugin_file in
pedroalvarezradiofree: I don't know. I will do the effort of reflashing and installing the jetsons if I can sort out the network limitations11:42
bashrcso I think the firehose plugin preamble is beyond the ken of inspect.getmembers11:44
KinnisonI've never seen python fail to understand firehose before11:45
Kinnisonesp. since what you linked to is what I wrote and was using for quite some time as a test platform11:45
bashrcone difference might be the presence of "from" imports and also the contextmanager stuff11:45
KinnisonI wrote that code and used it day-to-day for a couple of weeks11:46
KinnisonI doubt it has bitrotted that badly in the meantime11:46
radiofreepedroalvarez: maybe codethink could host it in the offices until that's sorted out?11:46
radiofreethis kills the upload11:48
bashrcI'll need to investigate more to find exactly what the bug is, but for now it looks like inspect.getmembers fails11:48
bashrcinspect.getmembers does find some stuff, but the wrong stuff in the plugin header11:50
Kinnisoncan you please pastebin how you are attempting to run firehose, and the full exception and backtrace?11:50
radiofreepedroalvarez: is it ok for me to ask datacentred on the list?11:50
radiofreeis anyone from there on the list? or on here?11:51
richard_mawyeah, since "PYTHONPATH=`pwd` MORPH_PLUGIN_PATH=`pwd`/plugin morph help" lists the firehose subcommand for me11:51
richard_mawwhen run from the subdirectory11:51
richard_maws/subdirectory/clone of the firehose plugin at the ref you gave11:51
KinnisonThere's even a convenient morph wrapper in that repo11:51
pedroalvarezradiofree: i'll msg you the details to contact them11:52
* Kinnison has no idea what is11:57
Kinnisonit's certainly not in the tree you linked to earlier11:57 is a script which runs morph with a firehose subcommand11:57 is my item, sorry11:58
KinnisonThen it'd help to know what it's doing11:58
perryl_it acted as a simplification of the command 'morph firehose item.yaml' to run as 'firehose item.yaml'12:00
Kinnisoncan you please use in future, it loads way faster than 0bin pastes for me12:00
*** mdunford has quit IRC12:04
KinnisonRight, so bashrc isn't using the code he referenced at
* Kinnison lunches and might be able to offer help afterwards12:05
bashrcby process of elimination I deduce that cliapp or inspect.getmembers gets befuddled by "from"12:07
KinnisonGiven that 'from' is normal python, I think you're barking up a bizarre and very wrong tree12:07
* Kinnison -> lunch12:08
*** edcragg has quit IRC12:08
*** edcragg has joined #baserock12:09
paulsherwoodbaserockers, is part of the g.b.o trove, or something else?12:11
pedroalvarezsomething else12:11
bashrcinspect.getmembers is really simple
pedroalvarezpaulsherwood: currently codethink is  hosting it12:12
radiofreewhere's that being hosted?12:12
bashrcok, it's the contextmanager. With that removed inspect.getmembers parses correctly12:18
*** perryl_ is now known as perryl12:18
bashrcI've no idea what the contextmanager does, but it's a troublemaker12:19
*** mdunford has joined #baserock12:20
pedroalvarezI thought that context-managers were awesome12:22
bashrcthey might be, but inspect doesn't like them12:23
paulsherwoodbashrc: a couple of simple examples -
paulsherwoodit allows you to for example set some stuff up, hand over, then unset it when your thing finishes12:26
ZaraSotK: Okay, I fixed it by pushing my branch to a public repo, then using morph checkout [branch] to get it back again. :P suggests error was down to morph not paying attention to git.12:27
radiofreepedroalvarez: so i think we've got plan for the arm mason12:28
radiofreei) create distbuild images with fail2ban ii) reflash them using new tools iii) setup authorized_keys, disable password authentication over ssh iv) setup fail2ban to instantly ban anyone that tries to ssh in with a password12:29
pedroalvarezand v) give the jetsons public IPs?12:31
pedroalvarezI thought that was the main problem12:31
radiofreei was under the impression that isn't a problem?12:31
radiofree(after a discussion outside of this channel)12:31
pedroalvarezradiofree: you mean that they can have public Ips?12:32
radiofreewell let's cross that bridge when we come to it12:32
pedroalvarezyeah but I didn't want to start building the bridge before knowing that is a river that we can cross12:32
pedroalvarezno need to integrate fail2ban if they are not going to have public Ips (for example)12:33
DavePagepedroalvarez: I'd still advise it TBH12:33
DavePageThere's very little downside and quite a lot of up.12:34
radiofreeit probably should be a default on distbuild networks anyway12:35
pedroalvarezyeah I agree, but disabling password authentication would be enough to avoid attacks12:35
pedroalvarezsecurity experts in 3...12:35
radiofreeyou can setup fail2ban to monitor things other than ssh logins12:36
pedroalvarezI agree, but if they are not going to have public IPs there is not point ( i believe)12:36
pedroalvarezIt's ok, is not a big deal, and is something we will do. I just would like to know if our network "limitations" in distbuild can be sorted12:37
radiofreei think the intention is they *will* have a public ip, it'll be the same setup as the last one12:38
DavePageI think that's mostly a qustion for Baserock Ops :)12:38
radiofreewho are they?12:38
pedroalvarezit is written down here:
pedroalvarezcurrently gary_perkins franred ssam2  and pedroalvarez12:39
radiofreeok, so you12:39
DavePageI'm just advising that fail2ban by default is sensible, because otherwise one day you'll forget to add it by hand.12:41
pedroalvarezI'll continue with its integration as soon as I can :)12:48
pedroalvarezradiofree DavePage: regarding Jetsons  having public Ips, I don't know. i contacted DC and i had really complicated answers, that's why I asked for help.12:49
radiofreepedroalvarez: fwd it to the list?12:50
pedroalvarez<bilco105> pedroalvarez: they'll all have internal IPs, which then NAT to a single IP address externally12:56
pedroalvarez<bilco105> i.e. all the jetsons will be behind one IP address12:56
richard_mawbashrc, pedroalvarez: Context managers are awesome, but they're supposed to be used as a decorator for functions, not classes, as a class' constructor should not yield12:58
richard_mawthe issue is that the last change to the firehose_plugin file removed the function the contextmanager was bound to, without removing the decorator, so the class ended up being given the decorator13:01
richard_mawbashrc: if, when I asked you for a link to the code, you had given me the link to the version you were using, I could have pointed that out an hour ago13:02
*** mdunford has quit IRC13:02
KinnisonIs it possible to submit a change to gerrit *without* git?13:11
*** mdunford has joined #baserock13:14
pedroalvarezgithub style? I don't think so13:16
KinnisonCan I push to gerrit without ssh support?13:17
pedroalvarezlooks like it's possble using http13:19
* Kinnison hmms13:19
* Kinnison waits for gerrit to generate an http password13:20
Kinnisonis this hard?13:20
*** mdunford has quit IRC13:20
Kinnisonapparently so13:20
pedroalvarezsome people had problems when doing it13:21
pedroalvarezI didn't13:21
Kinnison"no new changes" says gerrit13:22
Kinnisondo I *have* to have a changeid thing?13:22
Kinnisonbecause even the 'anonymous http' option tries to use ssh to get the commit hook13:22
Kinnisonis the hook unique per user, or could someone pastebin it?13:23
SotKit can probably be pasted13:25
SotKone sec13:26
SotKyou can probably add Change-id manually I imagine too13:26
KinnisonI don't want to risk it in case it encodes anything in there like UUIDs do13:26
pedroalvarezmadre mia13:27
pedroalvarez(wrong combinations of keys in the wrong window)13:27
KinnisonWoot, patch submitted13:30
jonathanmawIs there a good way of testing (or checking) whether the cache server specificed in artifact-cache-server is working?13:32
jonathanmawI'd like to know whether I've misconfigured morph.conf, or that the cache server doesn't have any artifacts for that.13:32
* Kinnison strongly suggests that someone test-build with his patch13:34
Kinnisonbecause I've only tested it with a non-morph build tool13:34
*** mdunford has joined #baserock13:36
paulsherwoodjonathanmaw: try parking your branch, then git fetch, git reset --hard master, then build, should pull some cachestuffs?14:07
jonathanmawnope, still attempting to build glib for system genivi-baseline-system-armv7lhf-jetson.14:10
SotKare you trying to use
jonathanmawSotK: no, ct-mcr-114:12
SotKwhats the cachekey its calculated?14:12
jonathanmawSotK: I see this
jonathanmawI'm not sure which one is the cache key14:15
SotKits probably mentioned in your morph log (probably /src/morph.log)14:15
SotKalternatively, doing `morph list-artifacts . HEAD systems/genivi-baseline-system-armv7lhf-jet14:16
jonathanmawSotK: glib seems to have cache key 21b5d7ea2f5e70d6606895edb0cd622a930f8884524f127a50285ba10fde296114:17
SotKOK, something has gone weird with the cache then, because the artifacts for that cache key are on ct-mcr-114:18
jonathanmawMaybe my configuration is wrong14:18
jonathanmawI have "artifact-cache-server = http://ct-mcr-1:8080/"14:19
jonathanmawin my morph.conf14:19
SotKcan you ping ct-mcr-1 from this machine?14:19
jonathanmawapparently not.14:19
SotKtry using its full domain name14:20
ZaraI've also been having slow builds recently, wondering if there's something up with the cache settings14:20
jonathanmawta SotK14:22
*** a1exhughe5 has quit IRC14:22
SotKno problem14:27
*** a1exhughe5 has joined #baserock14:40
pedroalvarezIs people using still feeling like this:
jonathanmawpedroalvarez: less so now I've run out of things that were already cached14:47
rjekObviously we need to upload cache items for every possible combination of software.14:49
*** zoli__ has quit IRC14:49
pedroalvarezfinally mason-x86_64 has finished building15:05
bashrcevery possible combination sounds combinatorially intractable15:09
radiofreeout of interest why is the build time in seconds?15:15
radiofreeand why did it take over 12 hours?15:16
rjekShould it be in light planck length seconds?  That way you can always represent it as an integer.15:16
radiofreesomething a bit more human?15:17
* rjek agrees it could be more human-readable15:18
radiofreei don't think i've ever heard someone say "wow, that took over 43,200 seconds!"15:18
DavePagerjek: I doubt you need granularity of more than a second for build timing15:18
rjekDavePage: :)15:19
ssam2jjardon: did you just get a mail from  ?15:20
ssam2it tried to send you one, but the Exim I just set up seemed to get an error from because it tried to lookup the mail address and didn't find it ...15:20 should work15:21
rjekAnd I can tell you that no attempt has been made to email it today.15:22
ssam22015-03-26 15:18:13 1Yb9Xu-0000dz-SZ ** R=dnslookup T=remote_smtp: SMTP error from remote mail server after RCPT TO:<>: host []: 504 5.5.2 <mail>: Helo command rejected: need fully-qualified hostname15:22
ssam2initial searching shows it may be a reverse DNS issue..15:22
rjekssam2: No15:22
Kinnisonssam2: Your MTA needs to send a FQDN15:22
rjekssam2: It means your Exim is sending "HELO foo", where "foo" is not an FDQN15:22
Kinnisonssam2: Your MTA is saying "HELO mail" or "EHLO mail"15:22
ssam2ah, yes I probably didn't configure that15:23
Kinnisonssam2: what MTA are you using? Exim?15:23
Kinnisonset primary_hostname appropriately in exim15:23
Kinnisonthat should solve the issue15:23
ssam2it's just using the hostname of the system,. which is 'mail'15:23
jjardonssam2: let me check15:23
Kinnisone.g. on our system whose hostname is 'platypus' we have:15:23
Kinnisonprimary_hostname      = platypus.pepperfish.net15:23
rjekssam2: Could you perhaps paste your exim.conf?  There may be other issues we can catch too15:25
Kinnisonssam2: FYI, both rjek and I run large exim setups :)15:25
radiofreepedroalvarez: do you think it would be better to build the systems in mason separately?15:26
radiofreeyou end up building every twice when these build-essential changes happen15:27
radiofreeif you build devel-system, then weston-system, wouldn't most of it be cached?15:27
SotKsurely it would be in the mason's local cache anyway?15:28
ssam2rjek, Kinnison:
Kinnisonprimary_hostname = baserock.org15:28
Kinnisonthat's a bit of a lie15:28
ssam2i've made minimal changes from the Fedora default config. I did it with the Ansible lineinfile module which is a bit ugly but means that I don't have to keep a copy of that huge default config somewhere15:28
rjekprimary_hostname should ideally be the external name of the machine sending the mail15:28
Kinnisonsurely the primary_hostname should be ?15:29
rjekie, gerrit.baserock.org15:29
ssam2it doesn't have one, it's not accesible from outside15:29
Kinnisonif that's where exim is15:29
Kinnisonssam2: Oh15:29
radiofreepedroalvarez: ah, it's actually doing that already?15:29
ssam2it's not, its in an isolated instance15:29
radiofree2015-03-26 03:37:01 Started building ccache-doc on mason-x86-64:343415:29
radiofree2015-03-26 03:37:01 Started building ccache-doc on mason-x86-64:343415:29
radiofreeso i suppose it's only building the thing once15:29
ssam2I thought that was sensible to stop it being hijacked?15:29
Kinnisonwell, it might be acceptable15:29
Kinnisonsome mail systems might refuse mail from it though15:29
rjekssam2: Ah, you're configuring a smarthost for use on the private side of the network, rather than on the machine sending email?  OK.15:29
rjekBut primary_hostname should then match the IP address that outgoing connections are NATed on15:30
* bashrc also runs exim15:30
ssam2ah, OK. That's not too hard to find15:30
ssam2although it may change... might be easier to give the machine a public IP and use firewall rules to prevent anyone getting to it15:30
ssam2or would that default the point as well?15:31
rjekssam2: You can control which interfaces Exim listens for SMTP on.15:31
rjekSo you just set it to only listen on the "internal" network15:31
rjekAnd/or only accept unauthenticated non-local mail from the relevent networks15:31
ssam2right. so the primary_hostname has to be something that can be pinged (pung?), but doesn't have to respond to SMTP15:31
ssam2is that right?15:32
Kinnisonit must resolve properly15:32
Kinnisonand the resolution must match the IP15:32
rjekssam2: It must resolve to the same IP address as it will be connecting out on15:32
ssam2I see.15:32
Kinnisonotherwise it doesn't need to accept mail15:32
rjekssam2: Is there a reason you're not running Exim *on*, and having it only listen on localhost?15:32
jjardonssam2: no email sorry (I checked SPAM folder as well)15:33
ssam2rjek: several people advised me to set up a separate relay15:33
rjekjjardon: GNOME's mail server didn't even accept the HELO :)15:34
ssam2rjek: there are more systems that will send email, already does, and hopefully storyboard will in future15:34
rjekssam2: OK, if wants a smarthost for all their servers, that probably makes sense15:34
ssam2i was going to add sSMTP to the gerrit system, but it seems fine to just set sendemail.smptServer to the IP of the 'smarthost'mail15:34
ssam2in the Gerrit system15:34
rjekBut I would at that point recommend that it has a real external IP address that is static and has bi-directional DNS agreement, and is configured to accept mail only from the netblocks that uses15:35
ssam2hmm. so we'd need to set up to point at its floating IP, essentially?15:35
ssam2that's doable I guess15:35
pedroalvareztiagogomes_: could you reviw this please?
rjekAlso, local_domains should probably be empty15:35
rjekAs it does not accept mail for anything15:35
ssam2ok, I misunderstood the meaning of that I guess.15:36
tlsapedroalvarez: I'm about to push an update with your suggestion15:36
rjekExim config is not simple :)15:36
tlsapedroalvarez: after I test building15:36
* rjek has spent many many many hours doing this and is still rubbish at it15:36
tiagogomes_pedroalvarez sure15:37
tlsaoh wait, morph's starting from 5/26215:37
pedroalvareztlsa: I made another comment regarding the pre-configure-commands15:38
rjekssam2: Otherwise that exim.conf looks fine to me.15:38
rjekssam2: The reason you had to disable the sender verify is that it thought *it* was responsible for, and it knew gerrit@ didn't exist.15:39
*** zoli__ has joined #baserock15:39
ssam2rjek: I see. any idea how I configure what interfaces it listens on?15:39
ssam2the firewall rules are blocking external traffic on port 25 anyway, so not super important15:39
*** locallycompact has joined #baserock15:40
rjekssam2: local_interfaces15:40
tlsapedroalvarez: I replied15:40
pedroalvareztlsa: I see, then I think we need to test your patch in aarch6415:41
tiagogomes_I can't see properly the comments in gerrit, it is just me?15:41
tlsapedroalvarez: without removing the preconfigure there was some kind of syntax error in the ./configure, I think15:42
tlsatiagogomes_: click on them to expand them?15:42
rjekssam2: You probably want something like: local_interfaces = <; ::1;; 192.168.222.X15:45
radiofreeoh right jjardon, you were saying 3.0 to mean "set the min version of the kernel to 3.0"15:46
rjekWhere 192.168.222.X is the internal IP15:46
radiofreejjardon: i'm fine with that15:46
tiagogomes_I was about to moan that moving nano out from core would mean that no editor would be available in a cross bootstrap system, but there is still busybox's vi15:47
tiagogomes_I always struggle with vi15:48
tlsamy reason for moving it was stratum level build deps make me sad15:48
tlsaand morph wanting to rebuild everything every time I tweaked the nano chunk morph15:49
tlsanothing build-deps on devtools stratum15:49
jjardonIf its not a chunk needed to build the other strata, Id say it should be out of core15:50
tiagogomes_Ok I saved a comment, but it appears as draft. Is this normal15:52
SotKyeah, you need to go back to the main change page and click "Reply..." to finalise them15:53
tiagogomes_SotK, ta15:53
pedroalvareztiagogomes_: then you are not worried about it not building on aarch64?15:54
tiagogomes_pedroalvarez, nano 2.4.0 already seems to support aarch6415:57
tiagogomes_of course it would be nice to test the change, we never know15:57
edcragghi, i'm having trouble pushing to gerrit, says "unpack failed: error missing tree"
tlsaedcragg: try with --no-thin15:59
edcraggahh, that's on the wiki, isn't it...15:59
jonathanmawhrm, qtwebkit requires ruby-1.9, and for some reason I can't get it to build.16:01
ssam2can you use ruby-2.0 instead?16:02
ssam2that builds16:02
ssam2and is in the ruby stratum already, so you can just build-depend on that stratum16:02
jonathanmawok, I'll try that.16:02
edcraggtlsa: yes that works, thanks16:02
*** mdunford has quit IRC16:05
*** mdunford has joined #baserock16:20
*** mdizzle has quit IRC16:27
*** mdunford has quit IRC16:32
*** mdunford has joined #baserock16:32
*** sherm_ has quit IRC16:33
jonathanmawhrm, it seems chunk morphologies in source code take preference over chunk morphologies in strata.16:33
pedroalvarezI hope that's not true :/16:34
ssam2they don't16:34
jonathanmawah, I was looking at the wrong stratum16:34
tiagogomes_ah, gerrit sends emails as well16:36
pedroalvareznice uh?16:39
pedroalvarezthanks Sam! :)16:39
Kinnisonssam2: top banana!16:39
pedroalvarezwe know that some people is not getting email yet, though16:39
Kinnisonwell it'll take it a while to get to me fr.ex.16:40
KinnisonI have greylisting and a particularly strict MX16:40
* richard_maw just pushed gerrit changes 104, 106 and 10816:48
rdaleare there gerrit repos mirroring the lorried upstream repos?16:52
ssam2I think we need set up for everyone to be able to receive the mails, which needs to wait for one of the Codethink ops to update the DNS records16:52
ssam2rdale: nothing in delta/ is mirrored in Gerrit16:52
rdaleok, so i'll send a small patch to the mailing list for them then16:53
ssam2rdale: it's possible to do that, but in general changes for upstream projects should go through the upstream review process16:53
rdalemy patch is actually for mini-utils which i think is a baserock thing16:54
ssam2that should be in gerrit, then16:54
ssam2is it not?16:54
straycat15:29  radiofree$ 2015-03-26 03:37:01 Started building ccache-doc on mason-x86-64:343416:55
straycat15:29  radiofree$ 2015-03-26 03:37:01 Started building ccache-doc on mason-x86-64:343416:55
straycat15:29  radiofree$ so i suppose it's only building the thing once16:55
straycatradiofree, that looks like a distbuild bug, distbuild should report that it's already building ccache-doc the second time, unless someone has intentionally changed that16:56
straycatbut in this case the message is just duplicated16:56
straycatso that's still a bug, but not the same bug as i initially thought16:57
pedroalvarezif two initiators are building in the same distbuild network, should one of them receive a different message ?16:57
pedroalvarezI mean, I think the message is ok16:57
SotKpedroalvarez: the second initiator should see something like "ccache-doc is already being built"16:58
*** a1exhughe5 has quit IRC16:58
pedroalvarezbut it wasn't it just started16:58
radiofreeis mason-x86_64 quite slow?16:59
franredbashrc, did something go wrong with your latest patch sent to gerrit?16:59
pedroalvarezsome component finishes building, so the distbuild network starts building the next one, two initiatiors wanted the next one. Should one receive "this is already building"? Or should both reveive "this has just started building on X"?17:00
franredit shows some links in grey also you have 2 commits where you introduce the context manager and in the second you remove it...17:00
pedroalvarezradiofree: looks like it is a bit slow, but not slower than the 32b version17:00
straycatpedroalvarez, one should receive this is already building iirc17:00
straycatit's been a long time since i touched that code though17:00
radiofreepedroalvarez: 12 hours to build a weston system is more than a bit slow :\17:01
SotKhm, in that situation it looks like they should both get "started building"17:01
SotKat least, thats what the network I'm messing with does17:01
*** zoli__ has quit IRC17:01
pedroalvarezradiofree: that's why I don't want to add a qt system to the CI :p17:01
straycatpedroalvarez, oh actually, i think it might be a little more subtle than that, it's hard for me to remember the details now, but there's a sort of intermediate state where the first inititor's request creates a new job, but that job is not running yet17:02
straycatif the second initiator jumps in with its request at this point, then it's possible for both initiators to receive the "this has started building" message17:02
straycati emphasise that this is all based on memory from several months ago, so i may be wrong :)17:03
*** zoli__ has joined #baserock17:21
*** jonathanmaw has quit IRC17:27
*** zoli__ has quit IRC17:28
bashrchah. firehose ran, without crashing. Not saying it did anything important though.17:33
*** mdunford has quit IRC17:34
*** bashrc has quit IRC18:01
ssam2jjardon: i don't think you'll receive emails until we have a subdomain set up -- seems to reject the 'HELO' that it sends currently, maybe it's hoping for a domain name18:03
ssam2we keep getting 'message delivery failed' emails when gerrit tries to send you an email right now :)18:03
Kinnisonssam2: where is the DNS for managed?18:04
Kinnisonssam2: is it with the mail, or elsewhere?18:05
persiaI believe we rely on Codethink for that still.18:05
jjardonssam2: ok, I will  talk with GNOME sysadmins18:05
rjekIt's on Mythic Beasts.18:05
Kinnisonjjardon: It's not GNOME's fault18:05
rjekGNOME are entirely right in rejecting HELO
rjekHELO and EHLO should have an FQDN.18:06
Kinnisonssam2: what IP should have?18:06
jjardonah ok!18:06
* DavePage heads to Mythic Beasts18:06
Kinnisonsambishop: I've done it18:06
Kinnisonssam2: I've done it18:06
Zaradoes anyone know what might cause a baserock system to build and deploy, but install a package as version 0.0 rather than the version specified? this just happened to me and it was a bit weird.18:06
KinnisonDavePage: already done18:06
jjardonrjek: I do not understand at all what you are talking about :)18:06
Kinnisonjjardon: and that's why you're not a mail admin :)18:07
ssam2Kinnison: thanks18:07
DavePageKinnison: Ah cheers. ssam2's email arrived while I was dealing with leaky aircon18:07
ssam2Zara: what do you mean by 'install a package' ?18:07
KinnisonDavePage: I gave it a 60s TTL until we can be sure it's riht18:07
KinnisonDavePage: once we're satisfied you can up that to the 3600s default18:07
jjardonKinnison: indeed :)18:07
ssam2jjardon: could you comment on something in Gerrit now to test?18:08
Kinnisonit may take a -ve ttl timeout before it'll work18:08
Kinnisongnome's DNS could have cached the NXDOMAIN18:09
Zarassam2: well, install a python package in this case (build system is python-distutils), version is specified via commit ref and unpetrify ref (0.0.21)18:11
ssam2Zara: ok. Only morph pays attention to those fields18:11
ssam2in fact, even Morph ignores unpetrify-ref, it's only for humans18:11
ssam2zara: the version number you may be seeing probably comes from in the repo of the Python project18:12
*** zoli__ has joined #baserock18:12
ssam2or a file referenced by (setup.cfg, or something else -- there are a few ways that setting version numbers can work for Python packages)18:12
jjardonssam2: comment done18:12
ssam2thanks, i got the comment instead of a message rejection this time18:13
ssam2that's that job done, hopefully18:13
Zarassam2: it's weird because this hasn't happened before, and I've been building things containing this package for a while. I might try repeating the process tomorrow.18:15
ssam2are you using a different ref now?18:15
Zarano, it's the same, though it's a different branch18:15
Zarabut the instructions for this package are the same18:15
ssam2so it's a differentt sha1?18:15
Zaranot for this package, no18:15
ssam2ah, right18:15
ssam2something more weird then18:16
Zarayeah. :S I'll poke around tomorrow.18:17
radiofreetiagogomes_: 2015-03-26 15:57:39 [Build 86/259] [systemd] Elapsed time 00:24:0818:26
radiofreewhat was it on the moonshot18:27
*** zoli__ has quit IRC18:28
*** zoli__ has joined #baserock18:41
*** ssam2 has quit IRC18:43
straycatcan people avoid writing bs reviews on the series i sent, it was a test, thanks, -1ing one minor aspect of something without providing an overall or balanced review is not even slightly useful18:46
straycatjjardon, ^18:46
*** rdale has quit IRC18:46
straycatit's not meant to be like this, and i think i give up18:56
*** straycat has left #baserock18:57
radiofreegerrit has a review interface you can use over ssh, neat18:57
franredbashrc, bashrc_, you can not use gerrit as a temporary repository, it is not the same as we were using gbo - gerrit is only for final patches18:58
franreds/it is not the same/you can not use it in the same way/18:58
*** franred has quit IRC19:23
*** zoli__ has joined #baserock19:31
*** edcragg has quit IRC20:11
richard_mawradiofree: Some fo the slowness in the systemd build is that it's run with max-jobs: 1 because there was a bug that occasionally made it fail. The other is that we currently still build the tests20:54
radiofreerichard_maw: max-jobs was removed from systemd during the last upgrade21:02
radiofreesimilar time on mason21:06
*** zoli__ has quit IRC21:09
* rjek pokes his VMs22:06
* jjardon just checked and has received an email from gerrit at 19:08 \o/22:14
jjardonThanks to the one that fixed the problem! :)22:15
rjekjjardon: yay! \o/22:15
*** gfinney has quit IRC23:29

Generated by 2.15.3 by Marius Gedminas - find it at!