IRC logs for #baserock for Thursday, 2015-07-09

*** zoli___ has joined #baserock04:45
*** zoli__ has quit IRC04:45
*** zoli___ has quit IRC06:29
*** zoli__ has joined #baserock06:59
*** mike has joined #baserock07:07
*** mike is now known as Guest9810507:07
*** rdale has joined #baserock07:29
*** mariaderidder has joined #baserock07:48
paulsherwoodperryl: not sure whether it's relevant, but you can see the poc cache server i did at https://github.com/devcurmudgeon/ybd/blob/return-of-server-poc/server.py07:48
perrylpaulsherwood: thanks! i'll take a look at it now!07:48
paulsherwoodtrouble is i just tried it and it doesn't work out of the box.... so it may be more hassle than it's worth. ping me if you want me to explain what it's 'supposed' to do :)07:50
perrylargh :( still, it could be the missing piece that helps me get mine running!07:50
paulsherwoodnote that code for uploading is commented out... https://github.com/devcurmudgeon/ybd/blob/master/cache.py#L15507:51
paulsherwoodand the forking idea is probably no longer workable since ybd now has support for multi-instances which are themselves forks07:52
*** jonathanmaw has joined #baserock08:06
*** gary_perkins has joined #baserock08:28
*** ssam2 has joined #baserock08:45
*** ChanServ sets mode: +v ssam208:45
*** ratmice___ has quit IRC08:47
*** ratmice___ has joined #baserock08:47
*** flatmush has joined #baserock09:07
*** flatmush has quit IRC09:08
*** flatmush has joined #baserock09:12
*** lachlanmackenzie has joined #baserock09:17
pdarHeya baserock, Im trying to build a thing with my baserock but I get the following error: http://paste.baserock.org/pirolejubu10:22
pdaranyone know why this might be?10:22
KinnisonI imagine an unexpected syntax error in one of your morphology files10:23
Kinnisonhave you changed any since you last ran morph?10:23
pdarI have, but not the one its trying to build. Does morph like them all to be error free?10:29
SotKpdar: yes10:30
Kinnisonpdar: morph validates the full morphology set10:31
pdarhmm, I removed what I thought the offending morphology might be and it still spewed the same error.10:32
pdaroh, no, I found another syntactically questionable one10:34
Kinnison:-)10:34
paulsherwoodpdar: have you tried ybd on it?10:48
pdarpaulsherwood: I still havent tried ybd yet10:50
paulsherwoodpdar: you might find it gives different/useful error info (no promises, though)10:53
pdarpaulsherwood: perhaps I'll give it a try11:01
paulsherwoodpdar: if you do, edit ybd.conf and set base from 'ybd' to '/src/cache' - it should be able to re-use your morph git repos that way11:21
straycatssam2, thanks for merging that lorry11:30
straycatssam2, are we upgrading the trove today perchance?11:31
ssam2apparently the OpenSSL vuln isn't released til 1700 UK time11:32
ssam2so if it is something we need to update for, it won't actually happen til Fri anyway11:32
ssam2will try to merge that Lorry patch today though11:32
straycatok, i can sort the README now if you like?11:33
ssam2that'd be great11:35
straycatahh, looks like xmlto-tarball failed to lorry, i haven't checked but my guess would be the ssl cert issue i thought i had the other day, if that is the problem then it will be fixed when we upgrade the trove11:38
straycatactually, i might be wrong about that11:40
pedroalvarezURLError: <urlopen error [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed (_ssl.c:581)>11:40
pedroalvarezstraycat: lorried now11:44
straycatoh, what did you do to fix it?11:44
pedroalvarezexatly what you said it might be needed :)11:44
straycatokay, i'm not sure i get why that's fixing it, since as Kinnison pointed out it's just setting up existing certs, and we run update-ca-certificates at system-integration time11:45
pedroalvarezI was investigating that too11:46
straycatand yeah lorries fine on my newly deployed devel machine...11:47
pedroalvarezoh, I have an idea of why11:48
pedroalvarezlet me verify it11:48
pedroalvarezthing is that the trove was upgraded after we put `update-ca-certificates` at integration time11:50
pedroalvarezso, my guess is that something went wrong in the upgrade11:51
pedroalvarezOk, so I went to cache.baserock.org (because it should be in the same state as g.b.o was before I ran this command) and all the certs there are wrong11:56
pedroalvareze.i, all the symlinks in /etc/ssl/certs are broken11:57
pedroalvarezand some of them even pointing to "/ca-certificates.inst/usr/share/ca-certificates/..."11:58
straycatahh okay11:58
straycatif upgrading cache.baserock.org doesn't fix the links we should probably find out why then :)12:02
*** gary_perkins has quit IRC12:02
pedroalvarezyup, that's why I'm not going to touch cacche.b.o, to see what happens12:03
pedroalvarezso... the orig subvolume of the current system of cache.b.o was ok12:05
pedroalvarezand orig/run of the previous version are wrong12:06
pedroalvarezMy current guess is that our upgrades mechanism doesn't handle symlinks?12:07
*** gary_perkins has joined #baserock12:13
ssam2ouch12:24
ssam2so this could point to a bug in baserock-system-config-sync ?12:24
pedroalvarezwell, is not a bug if it's expected to behave that way12:32
pedroalvarezhttp://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/tbdiff.git/tree/baserock-system-config-sync/baserock-system-config-sync#n10512:33
pedroalvarez                # chose a symbolic link in this order12:33
pedroalvarez                # of preference: Vuser, V2, V112:33
rjekhttp://pastebin.com/bhH74d2y <-- OpenSSL12:50
rjekSummary: Not as bad as it could be; attackers can cause some checks on untrusted certificates to be bypassed.12:51
rjekI don't actually think that's a panic-causing lets-rebuild-everything-right-this-momemt thing for Baserock infrastructure12:51
straycatstill, it'd be nice to upgrade trove so we can lorry zips and then we can upgrade perl to 5.22 without needing to create a repo manually13:11
perrylpaulsherwood: i've managed to fix a couple of minor issues with ybd/server.py, however when i try and run it it gives the server error message and exits; are there any required arguments for it?13:17
ssam2straycat: i'd be tempted to just update lorry manually on git.baserock.org13:23
ssam2our upgrade mechanism works quite well, but it's not as convenient as doing that13:23
pedroalvarezhm... I think MANPAGER='less -R'  in is not right always: invalid option -- 'R'13:24
pedroalvarezin etc/profile*13:24
pedroalvarezwhich is installed by install-essential-files13:24
*** mdunford has joined #baserock13:28
straycatssam2, ok13:32
Kinnisonpedroalvarez: -R is supported by full-featured less13:34
Kinnisonpedroalvarez: but perhaps not by busybox13:35
pedroalvarezKinnison: I confirm that it's not supported by busybox less. :)13:36
Kinnison:-)13:38
straycathrm, so having imported a module from cpan, the sha differences between the tarball that the import tool lorried and the tarball that my trove lorried mean i can't build my system, and now we have no unpetrify or petrify commands so as a user i seem to be stuck14:14
straycatso, can i go revert the commit that removed those commands now?14:18
KinnisonWhy are you stuck?14:19
straycatbecause i'm not about to manually update 30 shas14:22
KinnisonI see -- the way you said that implied there was 1 sha needing updating, not 3014:23
straycatapologies s/tarball/&s/g14:23
KinnisonDo your repos not have the old stuff in there anyway in tags?14:24
straycati don't understand the qestion14:25
straycat*question :)14:25
KinnisonWell, your issue is that a repo which was previously a tarball import and is now a CPAN import has different SHAs yes?14:25
straycatno14:26
Kinnisonoh14:26
Kinnisonthen I am confused14:26
straycatmy issue is that sha generated by the import tool differs from the sha on my trove, because the shas are apparently different for each tarball import of the same tar14:26
KinnisonSo you have local import repos14:27
Kinnison?14:27
straycatyes14:27
straycatthe import tool runs lorry locally to generate the definitions14:27
straycatbut you use the generated lorry files to get stuff into your trove14:28
KinnisonI see14:28
Kinnisonso tarball imports are not repeatable :(14:28
Kinnisonhmm14:28
straycatso what i want to do now is run unpetrify on this stratum14:28
straycatand then run petrify, so that i get all the correct shas from my duckhotel14:28
straycats/duckhotel/trove/14:28
KinnisonI see... Yes, that would be a use for that kind of functionality14:29
KinnisonHowever it's also sad that the local imports generated different SHAs14:29
* Kinnison wonders why14:29
straycatdifferent timestamps maybe?14:30
KinnisonI thought it was meant to stamp things properly, oh well14:30
straycatthe author timestamp is preserved, the commit timestamps seem to differ14:33
Kinnisonaah14:33
Kinnisonbummer14:33
straycatmaybe we could do tarball imports with libfaketime :314:37
Kinnisonperhaps14:37
* Kinnison imagines those involved with the bit-for-bit reproducibility work may be interested in helping with that14:38
* perryl perks up14:38
rjekYou've awoken perryl!14:38
Kinnisonruhroh14:38
perrylas long as libfaketime isn't doing anything relating to build mechanics, that could be a good use!14:38
* Kinnison has emperryl'd the channel14:38
perrylKinnison: :D14:39
* Kinnison will get his coat14:39
persiaThe other use case for repeatable lorries is to be able to exchange definitions between two users whose troves do not participate in the same hierarchy.15:10
*** rdale has quit IRC15:23
*** rdale has joined #baserock15:23
paulsherwoodperryl: what's the error?15:28
perrylSERVER ERROR: Something went wrong starting SocketServer15:29
perryli'm assuming it's something going wrong with the TCPServer, however there's no traceback info15:29
rdalemwilliams: Status "Go to Business Protection to be issued a Proximity Card"15:30
rdaleis that good?15:30
jmacsYes15:31
jmacsAlthough it's too late today15:31
KinnisonAre you chaps leaking?15:31
jmacsUm, maybe15:31
* Kinnison mops up15:31
paulsherwoodperryl: you need to set server to http://ip.address.of.your.machine:8000/ (for example) and create the /src/cache/remote directory (or change that to one that exists?)15:32
KinnisonGood error messages are hard to do15:33
Kinnisonas evidenced by this, and also by pdar's issues earlier15:33
perrylpaulsherwood: speaking of /src/cache/remote, are you accepting patches for server.py? i've added in a couple of lines that check if the directory exists, and creates it if not15:33
perryli don't believe server.py is in master so i assume not :)15:34
paulsherwoodperryl: i would happily accept a server.py that actually served :)15:35
perrylheh :) well, once it's working, i'll add in that check!15:36
straycatKinnison, the time can be set in the importer :)15:43
straycatthe commits now have the same commit timestamp but still different shas, so i guess there must be other factors15:47
Kinnisonaah well, you tried15:48
paulsherwoodstraycat: as a shortcut, can you use an old version of morph to unblock your work? but i agree this seems like a case for bringing back the commands15:49
paulsherwood(if i'm understanding what's happening correctly)15:50
straycatKinnison, possibly the file timestamps, those still differ15:53
straycatpaulsherwood, I could15:53
KinnisonI don't think file timestamps get into git15:53
Kinnisontree entries are type, content sha and leafname15:54
straycatokay15:54
perryl\o/ i've got ybd/server.py doing something15:57
KinnisonYou had it doing something this morning15:57
Kinnisonjust not a useful something15:57
perrylhaha15:57
perryls/something/something that isn't erroring/15:58
*** jonathanmaw has quit IRC16:07
*** gary_perkins has quit IRC16:14
*** zoli__ has quit IRC16:19
*** mariaderidder has quit IRC16:31
straycatoh, erm, i'm very sorry but this does work16:47
straycatmy test was, well, broken >.>16:47
paulsherwoodstraycat: that happens to us all from time to time :)16:48
straycati hadn't noticed that lorry puts the path into the commit msg >.> and i am being pretty stupid today anyway16:49
straycatokay i also think it would be sensible to modify the importer so that it just puts the filename in the commit msg rather than the full path to the file, since iirc the working area could be set differently on different troves17:01
*** rdale has quit IRC17:02
persia+1 to that idea17:02
ssam2yes, sounds like a good fix17:04
*** mdunford has quit IRC17:08
*** zoli__ has joined #baserock17:11
*** zoli__ has quit IRC17:14
*** zoli__ has joined #baserock17:17
*** mdunford has joined #baserock17:22
pdarHeya, Im testing a new strata, and dont really want to submit lorries for things just yet. This has worked ok for the git repos, which I just put in their repo URL into the appropriate field. I now need a tarball, is there an easy way I can do something similar for tarballs?17:22
straycatpdar, you can write a lorry file and run lorry locally17:23
straycatwith just $ lorry <lorry file>, then $ git clone workd/<lorry name>/git <lorry name>17:24
*** ssam2 has quit IRC17:26
*** lachlanmackenzie has quit IRC17:31
pdarthanks straycat, ill give that a go17:45
*** mdunford has quit IRC18:07
*** zoli__ has quit IRC18:32
*** zoli__ has joined #baserock18:36
*** zoli__ has quit IRC19:05
*** Guest98105 has quit IRC19:37

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!