[01:06] that issue of boot being stuck on /dev/sda1 is an annoying one that bit me in the past [01:06] and it's not KVM specific [01:07] it happens on baremetal too for me [01:07] and indeed it's just a vt problem, hit alt-f1 and you have the login prompt, so nothing is actually stuck [01:07] can't figure out even what promt that is, becaue I can't back to it with any of the F keys [01:12] drab: do you have a picture/video of the issue? === jstevewhite is now known as stwhite === stwhite is now known as jstevehite === jstevehite is now known as jstevewhite === JanC_ is now known as JanC === JanC_ is now known as JanC [09:38] xnox: why sftp? [09:39] rbasak, because it is encrypted, unlike annonymous ftp. And people should not be uploading security/embargoed/nda packages over unencrypted (default) channels. [09:45] Is that the only encrypted channel available? Is HTTPS available, for example? [09:53] nacc: I think that's in our roadmap? [09:53] (yes) [12:21] hi there! i am wondering which is the right config file for mariadb on 16.04.3 [12:22] there is conf.d as well as mariadb.conf.d [12:23] is it mariadb.conf.d/50-server.cnf ? [12:39] Frickelpit: mariadb.conf.d for MariaDB-specific configuration. In case you have a MySQL client installed as well, for example. [12:40] Though that combination will likely break things, separating the configuration allows you to switch more easily. [12:44] rbasak: no, i use postgresql but thanks for the hint ;) [12:46] Oh I'm sorry. [12:47] That was for friendlyguy, who has left. Autocomplete failure. [12:47] np :) [12:48] of course he left -you didnt repsond within 60s! [12:48] ok 120 [12:54] Never mind. I'm only the Debian MySQL maintainer who wrote the MySQL/MariaDB configuration file management mechanism for Debian and Ubuntu :) [13:11] Guys, how do I completely clean iptables (iptables-persistent)? [13:19] adac: if you want to stop applying any firewall rules, purging the iptables-persistent package and rebooting should do what you want [13:19] sdeziel, ok Trying to purge it then, thanks [13:20] adac: hmm, there is an easier way: service netfilter-persistent stop [13:21] (or iptables-persistent) [13:21] sdeziel, what happesn when I restart service iptables-persistent start again? [13:22] adac: it will bring the old rules back [13:22] adac: stopping the service purges the currently loaded rules and ACCEPT everything [13:22] sdeziel, hmm I see. Yeah I woulld like to have a command where I can erase these old rules from iptables-persstent [13:23] adac: like erase and start fresh or erase and get rid of iptables? [13:24] sdeziel, like completey wipe out all rules [13:24] adac: OK then stop the service and purge the package [13:25] sdeziel, kk Will do that. thank you! [13:25] np [14:45] rbasak: i meant it is in 'future', which to me is still nebulous :) [14:48] nacc: we could move it to 1.0 I suppose. I think we should do it soon, but whether it should be 1.0 or a 1.1 is debatable I think. [14:48] rbasak: yeah, it feels like the importer is the only 1.0 goal -- everythig else (cli changes, etc, which are independent of the hashes) are now get done whenever, but not blocking 1.0 [14:49] rbasak: or were we talking about different "it"s? [14:50] nacc: I agree [15:31] nacc: https://code.launchpad.net/~racb/usd-importer/+git/usd-importer/+ref/commit-graph-v2 [15:31] nacc: you want a resubmit MP for that? [15:32] rbasak: i think that would be good, as it will trigger a jenkins ru [15:33] OK I'll do it now. [15:33] rbasak: you may eed to rebase onto latest master? [15:33] *rebase [15:34] that doesn't affect the jenkins run, but will affect the meaing of the result and the actual merge :) [15:34] Oh [15:35] I did the MP already. [15:35] rbasak: it's ok [15:35] I can rebase and force push. Will that work? [15:35] https://code.launchpad.net/~racb/usd-importer/+git/usd-importer/+merge/332925 [15:35] rbasak: yep [15:35] rbasak: it might lead to two jenkins jobs, but that's ok [15:41] OK rebased and pushed. I had a few merge conflicts but I think I resolved them correctly. [15:42] rbasak: thanks [16:13] is it common to have fakeroot in the build-depends bit of d/control? [16:13] I have debhelper already, and everything else I need, but dpkg-buildpackage is complaining that fakeroot isn't installed [16:13] I thougnt it would be installed indirectly, but maybe I have been installing devscripts too often [16:13] this is on zesty [16:15] andreas: dpkg-dev recommends fakeroot [16:15] I have it, but didn't have fakeroot [16:16] debhelper pulled in dpkg-dev [16:16] andreas: i believe you can pass -r to tell it to use something other than fakeroot [16:16] which is why it's a recommends and not a depends [16:16] so every source package out there that does not have fakeroot in the build-depends list can only be built with a -r parameter? [16:17] or maybe our builders install recommends by default? [16:17] andreas: dunno [20:18] rharper: around? [20:18] is 'qemu-nbd -c' supposed to be reliable, i.e. not return until /dev/nbd0 is really available? [20:19] hallyn: [20:19] it's not [20:19] we had a bug in mount-image-callback related to that [20:19] some race between the kernel and the dm layer IIRC [20:22] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1628336 [20:22] Launchpad bug 1628336 in linux (Ubuntu Yakkety) "mount-image-callback cannot mount partitioned disk image" [Medium,Fix released] [20:22] some folks use udevadm settle inbetween qemu-nbd -c and whatever else they want to run [20:22] hallyn: === JanC_ is now known as JanC [20:28] rharper: is udevadm settle reliable for that? [20:28] if so, excellent [20:28] I think the blkdev reread in the bug is better [20:28] looking, [20:29] udevadm settle just pokes to see if any events need processing, where as the reread is more likely what's needed [20:29] rharper: so with that you think it woudl be reliable? Debating whether to spend the time doing loopback with offset mounts... [20:29] if nbd will ever fail then it's worth it, ... [20:29] why not kpartx ? [20:30] well i dunno, [20:30] you need nbd to do format to block device, then you need the kernel to probe for partitiions etc, and the kpartx will map the partitions to md (doing the offsets for you) [20:30] s/do format/translate format [20:30] have you looked at mount-image-callback ? [20:30] in cloud-image-utils [20:31] hm. [20:31] it's most certainly going to be what you want [20:31] lets you mount up images, and chroot to run stuff in them, it supporst bind mounting system mount points inside and copying in/setup/cleanup resolv.conf so you can apt-update, etc [20:31] i have LVM lvs on partition 2... [20:32] if blockdev --rereadpt /dev/nbd0 is going to work... [20:32] should be fine, kpartx will do the right thing; do your qemu-nbd, then kpartx -l /dev/nbd0; it'll show you the mounts it'll make [20:33] pretty sure m-i-c handles lvm inside as well [20:34] well, maybe just the kpartx; I can't say I've tried m-i-c with lvms inside; but kpartx will work for you (instead of your doing the offset mount yourself) [20:34] a veritable menu of options :) [20:35] so kpartx should end up faster than using nbd i assume [20:35] hallyn: "kpartx -s" will wait for the partitions to be created before returning [20:35] well, you need qemu-nbd if you have non-raw [20:35] if you have a raw then just kpartx [20:36] if you have qcow2 or vmdk or someother non-raw format that needs translated then you need qemu-nbd [20:37] just raw [20:37] thanks rharper and sdeziel ! [20:38] np [20:39] (actually i'll probably switch to qcow2 anyway to not have to worry about maintaining sparse file) [20:40] no!!! [20:40] i half expected that reaction [20:40] do you want your data or not? [20:40] well , you know , [20:40] kinda [20:40] =) [20:41] qcow2 eats data? [20:41] it's complicated but qcow2 is not a simple format; years ago we pushed on a simpler format that just did cow and a few other things [20:42] qed? [20:42] the metadata modification was large overhead to the format for high performance io; subsequently upstream qcow2 maintainers did a ton of work to clean up and improve performance which has made a huge diff [20:42] yea, qed [20:42] that pushed the buttons to get qcow2 into shape [20:42] but did qed ever get var enough to be usable? [20:43] absolutely [20:43] it was raw + cow [20:43] right i thought qed pushed qcow2 to be usable, and now qcow2 was acceptable [20:43] the format is still complicated [20:43] ok [20:43] thx for the info. Personally I only use zvols or LVs [20:44] well raw makes it easier to twiddle teh data later so i'll heed yoru warning and stick with that anyway [20:44] you can look at the internal coroutines needed to handle metadata writes; I'm an always raw person [20:44] just trying to decide whether to try figuring out how to use kpartx for this [20:44] kpartx -va /path/to/raw; mount /dev/mapper/loopNp3 /foo [20:44] or just use blockdev which is working . [20:44] umount /dev/mapper/loopNp3 ; kpartx -vd /path/to/raw [20:45] that sounds ideal [20:45] I think for lvms you get different names under mapper [20:45] but you'll see it [20:45] kpartx -l /path/to/raw will show you what it'll add without doing it [20:45] ideally i'd have an easily parsable output line to tell me which loop device to use; [20:45] but i'll play with it. [20:45] thanks again! [20:47] it does [20:48] it looks like it gives a lot of output, but i shoudl be able to do | grep loop | head -1 | awk ... [20:50] http://paste.ubuntu.com/25832415/ [20:51] that's some bash I had when parsing the output for picking out partitions [20:52] fmt is new to me [20:52] nice [20:55] rharper: but you're doing udevadm settle there? [20:56] hallyn: I'm pretty sure you can skip that by using kpartx -vas [20:59] yeah the manpage sure suggests so :) thx [21:00] yeah, old code; could/should have used -s === albech1 is now known as albech [21:38] hey got a quick question, i installed ubuntu server minimum server for vmware [21:38] i cannot seem to get a #!/bin/bash script working [21:39] and yes in the bash script it has sudo before the commands [21:39] has anyone had this problem before, and yes i did search [21:40] what errors are you getting? [21:41] DaddyEric: your statements don't seem to go together (logically, to me) [21:41] sudo: unable to execute ./multicraft-setup.sh: No such file or directory [21:41] Hangup [21:41] DaddyEric: care to pastebin the script? [21:41] DaddyEric: and presuambly the directory you are running it in's `ls -ahl`, if you are using relative executionn in your script [21:41] https://hastebin.com/gutecaheza.txt [21:41] yes it in fact is [21:42] DaddyEric: is this ... a hack to not use ansible or any number of other system management tools? [21:42] DaddyEric: 1) you are using /bin/sh, not /bin/bash [21:42] yes i changed that [21:42] and it still does not work [21:43] DaddyEric: 2) i see no mention of multicraft-setup.sh [21:43] not you're not using /bin/sh but bin/sh [21:43] sarnold: ha! :) [21:43] i know that i tried both [21:43] try changing #!bin/sh to #!/bin/sh [21:43] DaddyEric: dos it have execution flag? i.e. chmod +x [21:43] i would not be here if it didnt [21:44] DaddyEric: ok, no need to be short, you came here for help :) [21:44] DaddyEric: line 78 looks suspect [21:44] DaddyEric: can you pastebin the exact command you ran and the exact output [21:44] DaddyEric: change the 'sudo cd' to 'cd [21:44] DaddyEric: also line 73 [21:44] DaddyEric: line 73 too [21:44] I feel like this script seems pretty .. bad [21:44] * sarnold ^5s nacc [21:44] way too much sudo [21:44] aye [21:45] i do that in case the user perms does not have permission for that directory [21:45] it's better to just make the user run sudo ./script [21:45] yes and i have tried all of that [21:45] DaddyEric: still no invocation or exact output [21:45] i would not be here asking for help if i did not know how to do that [21:46] i try sudo ./multicraft-setup.sh [21:46] DaddyEric: also, consider using #!/bin/bash -x (or calling it with /bin/bash -x /path/to/script) so we can see what lie it's failig at [21:46] and i get the bad filename [21:46] nacc: good idea [21:46] DaddyEric: `ls -ahl .` [21:47] DaddyEric: it does't say 'bad filename', it was tryign to run bin/sh before, probably [21:47] DaddyEric: show us the *current* script and the exact execution output [21:47] it does not do the apt commands before it [21:47] DaddyEric: what? [21:48] DaddyEric: give the output requested. [21:48] https://hastebin.com/wavebahuxo.txt [21:48] $'update\r' [21:48] DaddyEric: and the exact script /home/dtsmc/multicraft-setup.sh [21:48] why is it doing that? [21:49] i already pasted that [21:49] DaddyEric: no, you did not [21:49] uless you left it at bin/sh [21:49] https://hastebin.com/icupabakoh.txt [21:49] you didn't happen to write this thing on a windows machine did you? [21:49] DaddyEric: seriously. [21:49] DaddyEric: ow it says #!bin/bash [21:49] DaddyEric: which is wrog [21:49] *wrong [21:49] it is copied exactly from a working script [21:49] DaddyEric: that script was wrong too [21:49] and yes i used notepad++ to write it [21:50] DaddyEric: i don't really care if it worked [21:50] DaddyEric: well don't do that [21:50] DaddyEric: `file /home/dtsmc/multicraft-setup.sh` [21:50] DaddyEric: it's probably got embedded windows line endings [21:51] multicraft-setup.sh: ASCII text, with CRLF line terminators [21:51] how do i remove them? [21:52] convert to asni right? [21:52] dos2unix, iirc [21:53] DaddyEric: and fix the shebang to be the correct path [21:55] multicraft-setup.sh: Bourne-Again shell script, ASCII text executable [21:55] thats the output from file [21:55] DaddyEric: oh, now provide the output from -x again [21:55] there we go simple fix [21:55] its working now [21:56] DaddyEric: ok [21:56] GRUB_CMDLINE_LINUX="consoleblank=0" [21:56] is that correct for disabling the console blanking