IRC Logs for #crux-devel Thursday, 2017-07-27

jaegerIf we fix it, sure :P00:00
teK_just fiddling with the subdir thingy00:01
teK_it promptly broke stuff00:01
teK_remote: FATAL: bad group '@__split_keys__'00:02
teK_how helpful00:02
teK_gitlab cant, apparently:
jaegersurprising; seems like that would be a popular feature00:03
teK_gitolite calls this namespace00:04
teK_This feature is only available in the 'namespaces' branch until enough people test it and tell me it works fine.00:04
teK_ namespaces Updated 5 years ago by sitaramc00:05
jaegerThat's not promising00:05
teK_diff is tiny, though:
jaegeryeah, not too bad00:07
jaegerAFK for a bit, need to eat00:08
teK_I need to sleep00:09
teK_take care and.. enjoy00:09
teK_gitolite aliases might save our day, though..00:13
pedjaRomster, take a look at FS#1442 when you have time (wxgtk update)00:15
teK_jaeger: git clone works now00:17
pedjaheh, I still get 'connection refused' for opt/john source00:18
pedjahttp works, https doesn't00:21
teK_oh and multiple keys at least gives warnings jaeger00:27
teK_see @split_keys above. This is you! :P00:27
teK_pedja: use our distfile mirrors00:27
pedjathat's an interesting idea. could 'pkgmk -do' be taught to use them as a fallback?00:30
jaegerteK_: ok, cool00:31
pedjaonly the cat and me are awake this late in my house00:33
*** onodera has quit IRC00:35
jaegerteK_: can I have permissions to clone the gitolite-admin repo next time you work on it?00:35
jaeger makes the multiple key thing seem really simple, using different directories per key00:40
jaegerIf you're already doing that, cool... if not, could do something like keydir/jaeger/{vindication,cygnus,intrepid}/jaeger.pub00:41
jaegerfrom the docs it seems this would work, if they're all called jaeger.pub00:42
jaegerfrom what I can see in ~git the way it's currently setup isn't quite like that, so if you have no objections, let's try it tomorrow00:51
*** _________mavric6 has quit IRC02:34
*** _________mavric6 has joined #crux-devel02:35
frinnstteK_: your vim stuff WAS annoying - it printed 40 lines of fuckin errors for the rest of us :)07:36
frinnstteK_, pedja: according to @solardiz they never had https on openwall.com07:41
teK_frinnst: should be fixed07:51
teK_also: Nerd-Swag? Really? :p07:55
frinnstI bought it from rms himself!08:01
teK_jaeger: sure, will also add frinnst and jue, obviously. Sepen, too. If he manages to reappear08:54
teK_frinnst: objections against running ports -u on regularly?08:59
teK_same-ish for prt-get sysup?08:59
frinnstnope, thats what we want to do. Not allow it to become like the old server :-)09:43
frinnstdid you see the images of the old server?09:44
frinnstnot sure where i posted those..09:44
teK_arctic cooling <309:45
frinnstbtw charlie offered to mount another disk in the server for us to keep backups on09:45
teK_nice cut job with the server label, too09:45
teK_hot standby =)09:45
teK_poor man's09:46
frinnstthey used to take backups of the old server (well, until the backup disk died). He asked if we wanted that on the new server too09:46
frinnstbut I said it wasnt necessary since jaeger runs his backups and I plan on doing the same09:46
frinnstdont want to leech off him too much :)09:47
teK_fully agreed09:48
teK_he has been very good with us09:48
frinnstand I offered us all to help if they have any linux issues they cant work out. questions and the like09:48
teK_minus the systemd things (at least for me) :P09:49
frinnsthehe yeah09:49
frinnstthey are almost all debian09:49
frinnstI am worried about the battery on the new server tho. When i rebooted a new kernel the server didnt come back up. Turns out it stopped during post because a mouse wasnt connected. Charlie fixed that09:51
frinnstI had rebooted it lots of times without that issue when we hosted it09:52
teK_nice catch09:52
teK_shall we send him a new one? :D09:52
teK_also: pictures of the new server available?09:52
frinnstno, he didnt replace the battery - just disabled the mouse/keyboard error thingy09:52
teK_Ill be right back09:53
frinnstit might also be because we actually had a keyboard connected but no mouse. it booted right up when we started it up without a keyboard09:53
frinnst <- from when it was hosted at my work09:56
teK_nicer label job10:16
teK_toying around with mirrorbits, btw11:42
teK_looks good :]11:42
*** onodera has joined #crux-devel13:14
*** darfo has joined #crux-devel14:50
*** nthwyatt has joined #crux-devel14:59
*** darfo has quit IRC15:01
*** chinarulezzz has quit IRC15:30
*** darfo has joined #crux-devel15:44
*** nthwyatt has quit IRC15:45
*** nthwyatt has joined #crux-devel16:22
*** darfo has quit IRC16:25
*** nthwyatt has joined #crux-devel17:17
*** darfo has joined #crux-devel17:23
*** nthwyatt has quit IRC17:24
*** nthwyatt has joined #crux-devel17:24
*** darfo has quit IRC17:25
*** nthwyatt has quit IRC17:25
*** darfo has joined #crux-devel17:25
*** nthwyatt has joined #crux-devel17:25
*** darfo has quit IRC17:25
*** darfo has joined #crux-devel17:30
*** nthwyatt has quit IRC17:30
*** nthwyatt has joined #crux-devel17:31
*** darfo has quit IRC17:35
just_funpedja, for 'pkgmk -do' with fallback distfiles see
*** nthwyatt has joined #crux-devel18:58
*** Workster has quit IRC19:12
*** nthwyatt has joined #crux-devel19:20
*** darfo- has joined #crux-devel19:21
*** nthwyatt has quit IRC19:24
*** darfo has joined #crux-devel19:24
*** darfo- has quit IRC19:29
*** pedja has quit IRC19:48
*** chinarulezzz has joined #crux-devel20:17
*** pedja has joined #crux-devel20:37
teK_mirrorbits looked good but passed rsync:// url to rsync, which interpreted that as NON rsyncd connection, disallowing another parameter and thus mirrorbits failing :O21:14
jaegerteK_: what do you think of the keydir organization I suggested for gitolite? seem reasonable?21:17
teK_what do you think about the one already in present? "}21:17
teK_let me check, did I miss that21:18
jaegerwell, you complained about the current one because it threw warnings about the split key group :P21:18
jaegerSo I suggested something different based on the documentation21:18
teK_I picked folder = group21:18
teK_did you check that out?21:18
teK_literally :o21:19
jaegerI still can't checkout the gitolite-admin repo but I can see what was done on the filesystem21:19
teK_may be because the multi key stuff is broken21:19
teK_let me quickly finish with mirrorbits21:19
teK_just another ~30m I hope ...21:19
jaegerPersonally I don't think the filesystem organization needs to match groups, considering the way they do multiple keys21:20
jaegersure, no rush, I was just curious21:20
teK_watch me failing at
jaegerhrmm, what does this do?21:23
jaegerreplacement for your distfiles script?21:23
teK_and mirror health checking21:23
teK_thought to give it a try21:24
teK_oh wow, the dot on the map went green21:24
teK_oh and it does geoip based routing21:24
teK_which should be good for our beloved users :-]21:24
jaegersounds good21:25
teK_it's written in go, though21:26
teK_got it to work :}21:37
jaegergood deal21:39
pedjaserverop is in the middle of the lake, it seems :)21:39
teK_so that's the ports repo only which I seem to be the only serving it via HTTP ;-)21:39
teK_for improved cooling21:40
pedjanice :)21:40
teK_also, afaik, it's hosted in N├╝rnberg ;)21:40
teK_we can have another instance for the disfiles. Who is running those mirrors as of now?21:41
teK_Romster, me .. frinnst also?21:41
jaegerI've got one as well21:41
teK_the only thing about mirrorbits is, that it also needs to have the files at the source machine = crux.nu21:43
teK_not serving but still present21:44
teK_meaning we would have to have a somewhat more centralized approach ( need not be the master, though)21:44
pedjateK_, I have a script that updates geoip data, if you are interested21:46
teK_there was one in contrib ;>21:46
teK_so mirrorbit's contrib/21:47
teK_but thanks!21:47
pedjaI took it from debian, iirc :)21:47
pedjamirrorbits is similar to mirrorbrain, I see, which is way more complex to set up21:52
teK_well I had to change the source and recompile it to make it work with our version of rsync(1)21:53
teK_also mirrorbrains -debug -D is a lie21:53
pedjabut it's still a single binary, which is cool21:54
pedjahow is our version of rsync different?21:55
teK_it's current21:55
pedjaah :)21:55
teK_mirrorbits needs a config and three template HTML files, next to the binary itself21:55
teK_libgeoip is the only runtime-dependency21:55
teK_frinnst: who is in charge of crux' DNS? mav still?22:02
teK_oh and best part?22:04
teK_mirrorbits depends on redis.22:04
teK_second use case of redis for us within three days22:05
jaegerwoot, heh22:08
teK_jaeger: where would you suggest the iso files go on
teK_atm they're in ~tek/repo22:09
jaegerhow about /var/www/htdocs/files somewhere?22:09
teK_sure. include or exclude from backup?22:10
teK_you have to check if the 23GB extra are OK22:10
jaegerI've got plenty of room locally22:10
teK_btw they do not have to lie within $WWWROOT22:10 is not supposed to serve them22:10
jaegerAh, ok, didn't catch that part22:11
jaegerMaybe in /srv/mirror or something similar?22:11
teK_we don't want to ruin mav's bw bill22:11
teK_will put them there, once rsync is done22:11
teK_would be cool if one could run a lame duck-style version without the fat files on mirrorbits22:13
teK_but I guess, one could do even that with disabled local rescanning and then updating redis from the outside, master source22:13
teK_because redis just holds all the hash sums for the files22:13
pedja'lame duck-style'?22:17
teK_now, it needs to actually hold the full content22:17
teK_it periodically scans the content and stores the checksums22:17
teK_ / is too small for /srv/mirror22:38
*** __________mavric has joined #crux-devel22:41
*** onodera_ has joined #crux-devel22:41
*** _________mavric6 has quit IRC22:41
*** heroux has joined #crux-devel22:42
*** onodera has quit IRC22:43
*** heroux has quit IRC22:45
*** heroux has joined #crux-devel22:45
*** onodera_ has quit IRC22:51
*** Workster has joined #crux-devel23:23

Generated by 2.14.0 by Marius Gedminas - find it at!