=== stevenroose|BNC is now known as stevenroose [01:25] hi, what precautions should be taken for HD crash in server? [01:25] jamespage, coreycb: aodh is not working on mitaka: 2016-01-12 01:20:27.950 | aodh-api : Depends: q-text-as-data but it is not installable / https://jenkins04.openstack.org/job/gate-puppet-openstack-integration-scenario001-tempest-dsvm-trusty/92/console [01:26] I mean, what are suggested backup policy for such accident . [01:27] jamespage, coreycb: https://bugs.launchpad.net/cloud-archive/+bug/1533014 [01:27] Launchpad bug 1533014 in Ubuntu Cloud Archive "aodh can't be installed" [Undecided,New] [01:29] cool_boy: there's two usual approaches: (a) use something like zfs or mdadm or hardware raid cards (b) be sure to have backups onto external storage, whether another server or some internet-based backup echanism [01:30] cool_boy: there's multiple risks to mitigate against: (a) bitrot on hard drives (b) dying hard drives (c) dead hard drives (d) dying / dead hard drive controllers (e) dying / dead motherboards, powersupplies, NICs (f) dead power source (g) dead network infrastructure (h) fire, flood, war, terrorism, etc [01:30] cool_boy: mitigate against all that you can afford [01:33] sarnold: thanks, so how backup onto external storage works? eg: I have an application running on server I take regular backup of db, but attachments are stored on fs [01:34] how attachment should be backup on external device, everytime full backup or sync server directory to external storage? [01:34] cool_boy: rsync is one option; I use rsnapshot to backup from one drive to another in my laptop, it provides some snapshots too; you can also use e.g. block-level backups such as zfs send [01:34] asking because attachments are like in 2 months attachments are like 600 MBs and certainly will grow with time [01:45] going to use rsync for now, will learn rsnapshot later [01:45] thanks a lot sarnold [01:45] have fun cool_boy :) [01:45] :) === stevenroose|BNC is now known as stevenroose [05:47] EmilienM, coreycb: backporting the missing bit now === ibaric_ is now known as ibaric === chmurifree is now known as chmuri [12:25] jamespage: thanks [12:26] jamespage: can you update https://bugs.launchpad.net/cloud-archive/+bug/1533014 when it's done? So I can re-run testing in our CI [12:26] Launchpad bug 1533014 in Ubuntu Cloud Archive "aodh can't be installed" [Undecided,New] [12:39] EmilienM, yuo [12:41] EmilienM, I've just promoted that package to -proposed - but it will take ~2 hrs or so to build and publish [12:43] jamespage: ack, thx [12:46] jamespage, EmilienM, that'll fix the install but then the aodh-api upstart script fails because it needs python-pymongo. I'll update the package for that. [12:46] coreycb, hmm ok [12:46] crappy [12:48] ok thx guys [12:48] jamespage, coreycb: FYI https://etherpad.openstack.org/p/puppet-openstack-ci-mitaka [12:49] if you want know ubuntu status in our CI [12:51] EmilienM, thanks for all the early testing === kickinz1 is now known as kickinz1|lunch [13:25] Hey, I have a ZTE MF823 3g dongle. When I plug it in, it shows up as enx364b50b7ef44 in ifconfig -a. What can I put in /etc/network/interfaces in order to bring up dongles like this automatically? - I tried "auto enx*" but it seems I cannot use a wild card. Any suggestions? [13:33] hackeron: enx364b50b7ef44 [13:34] hackeron: you may want to configure udev to give it a little more admin-friendly name [13:34] RoyK: I need it to work with any 3G dongle like this [13:34] RoyK: how? :) - that is the question essentially [13:36] hackeron: perhaps try to disable biosdevname - not sure how, though [13:36] that's the "consistent device name" troll [13:39] RoyK: so no way to use a wildcard or regex in /etc/network/interfaces? [13:45] hackeron: not to my knowledge. How would it know how to consistently find an interface? [13:45] hackeron: better to arrange for your interface name to be consistent. I believe it's quite configurable now. [13:46] hackeron: or alternatively depending on your use case you could arrange a "wildcard" udev rule to do what you want for any hotplug event. [13:47] rbasak: hmmm, I like "quite configurable" :) - where do you configure it, do you know? - trying to find on google also [13:56] jamespage, I attached a debdiff to Emilien's aodh bug if you wouldn't mind sponsoring that to aodh in ubuntu xenial === kickinz1|lunch is now known as kickinz1 [14:03] coreycb, ok - in a bit [14:11] hackeron: which release? [14:13] would some core dev have spare time to sponsor LP: #1522346 [14:13] Launchpad bug 1522346 in nut (Ubuntu) "Please merge nut 2.7.2-4 (main) from debian (unstable)" [Wishlist,Confirmed] https://launchpad.net/bugs/1522346 [14:14] this is one of my work item for 16.04 [14:18] caribou: I'll review it this week. Thank you for your work! [14:19] rbasak: thanks! I sorta fell off my radar :-) [14:51] rbasak: vivid [14:56] hackeron: then I think it's prior to the most recent change to ifnames. [14:56] hackeron: so just plain udev, or biosdevname if you've ended up with that enabled (IIRC it depends on your hardware) [14:56] But biosdevname won't apply to USB I don't think. [14:57] So you should be able to tweak it with plain udev. [14:57] See udev's documentation, /lib/udev/rules.d/ and /etc/udev/rules.d/ for the default interface naming rules. [16:09] stgraber: are the images on images.linuxcontainers.org regularly refreshed? [16:10] pmatulis: yep [16:10] pmatulis: usually daily, unless something's failing [16:11] stgraber: thanks [16:13] stgraber: are there instructions somewhere for setting up an image server? [16:14] (like a private one) [16:15] i don't suppose you can just point an lxd daemon at another one that has local images? [16:19] still trying to connect my verizon lte modem to an ubutu server via usb cable...does anyone know what packages are required for this to work? [16:21] You'll probably need to find the actual manufacturer and model [16:22] i got that info [16:24] Novatel Wireless [16:25] Different makes use different subsets of the AT Command set to turn them on and off, et [16:27] grendal_prime: This one? http://www.verizonwireless.com/dam/support/pdf/user_guide/u620-at-command-reference-guide-7-17-15.pdf [16:27] grendal_prime: you need the "pci id" - https://wiki.debian.org/HowToIdentifyADevice/PCI [16:29] na, its the jetpack one [16:30] hmm...ya its werid lspci it does not list this thing but lsusb does [16:32] unless its this guy... [16:32] 00:16.0 Communication controller: Intel Corporation 7 Series/C210 Series Chipset Family MEI Controller #1 (rev 04) [16:32] grendal_prime: Can you unplug it from the usb port, wait 30 seconds, plug it back in, wait 30 seconds more, then pastebin the results of: dmesg | tail -n30 [16:33] i can do that....i have some rope up here.. [16:34] sorry bad movie reference really. ahum...ya working on the dmesg info now [16:35] Hopefully we can find out from there what /dev it goes to, something like usually /dev/ttyUSB0 or /dev/ttyACM0 [16:36] which pastbin? [16:37] nevermind [16:37] http://paste.ubuntu.com/14478874/ [16:39] i ound a manual for that device reading now [16:39] Looks like it's mapping an ethX port to it [16:44] grendal_prime: What says result of: ifconfig eth0 [16:45] it comes up as eth5 [16:49] you know..its starting to look like this is more trouble than its worth. Im going to just throw a wireless usb nubby in there and just have it use that. to connect to the jetpack. [16:50] the desktop distros have little to no prolem connecting to this thing so i was hoping there was a way to just fire off some script to connect to it..but i dont see a way. [17:14] wow..looks like you can just use wvdial...it understands that your not using a real dial up modem..it even has a "stupid" mode [17:20] Yes, one of the advantages to the 3G/4G/etc AT command extensions is a regular old modem dialer can communicate with it [17:22] rbasak: https://git.launchpad.net/~nacc/+git/php7tracking/tree/package-list.rdepends.src.split.working [17:24] https://www.youtube.com/watch?v=A4urP-GlSwg [17:24] nice litle vidoe..I think im gonna need to use a physical machine to test it though... [17:25] usb host connection in libvirt...seems to be not so great.. [17:25] ill try it and see though [17:52] jamespage, coreycb: scenario002 is passing on trusty mitaka \o/ but scenario001 is still failing because of Aodh [17:52] http://logs.openstack.org/35/255635/16/check/gate-puppet-openstack-integration-scenario001-tempest-dsvm-trusty/f4ab95b/console.html#_2016-01-12_16_22_42_991 [18:01] arges: were you planning on doing another libvirt merge, or was that enough? :) [18:03] arges: also curious if you have any ideas on https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/1531833 [18:03] Launchpad bug 1531833 in libvirt (Ubuntu) "package libvirt-bin 1.2.16-2ubuntu11.15.10.1 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [High,New] [18:04] hallyn: I was going to discuss tomorrow about it [18:04] hallyn: i see that 1.3.x is in unstable [18:04] looking [18:04] * arges needs to fix the weechat indicator plugin... [18:05] so the unit goes into a failed state... not really sure why [18:07] hallyn: did they try restarting the service too many times? I see 'failed (Result: start-limit)' [18:08] oooha unit. that msg shows up all over the palce [18:09] *$#&)*$#)(*$ crappy network [18:09] arges: i assume it restarts [18:09] but actually, i [18:09] Hey! Is it normal my (pretty much stock) ubuntu server got 14 ram disks a 64MiB? [18:09] i'm thinking it' ssomething to do with their language pack, maybe. if you look at src/logging/log_daemon.c, it'll fail if you don't have that set [18:09] *15, havn't seen that before [18:10] hm, nah [18:10] oh, no, yeah, [18:10] arges: if you look at the strace the last thing it tries is language gorp [18:11] hallyn: hmm, so do we need ot patch log_daemon to fail more gracefully [18:11] allah is doing [18:11] sun is not doing allah is doing [18:11] arges: i'd say so [18:11] but we should first confirm :) [18:11] moon is not doing allah is doing [18:11] i really don't know what packages to check for [18:12] stars are not doing allah is doing [18:12] planets are not doing allah is doing [18:12] galaxies are not doing allah is doing [18:12] oceans are not doing allah is doing [18:13] mountains are not doing allah is doing [18:13] trees are not doing allah is doing [18:13] hallyn: language-pack-* i believe [18:13] mom is not doing allah is doing [18:13] dad is not doing allah is doing [18:14] boss is not doing allah is doing [18:14] dollar is not doing allah is doing [18:14] degree is not doing allah is doing [18:14] medicine is not doing allah is doing [18:14] customers are not doing allah is doing [18:14] !ot | guest_65854 [18:14] guest_65854: #ubuntu is the Ubuntu support channel, for all Ubuntu-related support questions. Please use #ubuntu-offtopic for other topics (though our !guidelines apply there too). Thanks! [18:14] sssssh [18:15] guest_65854 is not doing [18:15] guest_65854, please take your profanity elsewhere, thanks [18:15] you can not get a job without the permission of allah [18:15] !op [18:15] Help! Channel emergency! infinity, soren, lamont, mathiaz, Pici, Daviey, Tm_T, pmatulis, Corey, IdleOne, ikonia, funkyhat, Myrtti, ocean, genii, phunyguy! [18:16] Pici: noting this guy was doing the same thing on a different nick in #maas last night (no ops around there then) [18:16] so probably will be an issue elsewhere [18:16] rww: noted [18:17] thanks [18:17] felt like i had tinnitis for a second [18:21] Having 2 new harddrives installed, 4TB each, want to make a software raid with mdadm. Created a partition on each, I remember from the setup that the recommended partition type is "linux raid autodetect", but I cant find any article stating how to create a partition of that type? [18:22] Schalla: what are you using to create the partitions? [18:22] fdisk [18:22] Schalla: I believe you can ask fdisk to tell you the codes for all partition types [18:22] fdisk -> n -> (leave all defaults to create 4tb table) -> w [18:22] yeah, if you press "t" and then "L" [18:23] But that doesnt contain Linux raid autodetect, only linux raid [18:23] which is confusing [18:23] https://raid.wiki.kernel.org/index.php/Partition_Types ? [18:24] I don't think you need autodetect anymore, but not sure [18:24] Hm. [18:24] It used to be "fd", but that doesnt exist anymore. [18:25] i think you can just pick linux raid, but not sure [18:30] for some reason tho, after adding them to cups then sharing them, the cups server does not proxy the ipp requests i tells the clients to connect directly to the printer ip's [18:30] hi, trying to configure cups to proxy ipp. i have my ptrinters in a vlan and want to share them out via cups. [18:35] Schalla: don't use partitions if you want raid on them [18:35] Schalla: no need [18:36] Schalla: just mdadm --create --level=1 --raid-devices=2 /dev/md1 /dev/sdc /dev/sdd (replace with your own device names, obviously) [18:37] Schalla: partitons are there to partition the disk, as in split it up, if you don't want to split it up in partitions, no need for them for RAID to work [18:37] cleaner setup without them, too [18:42] RoyK: Thanks again. :D [18:43] Type fd is linux raid autodetect, and there's nothing wrong with using partitions, but fdisk uses MBR, which won't work on drives > ~2.2TB. You should be using a GPT tool, such as parted or gdisk. [18:43] qman__: or no pratitions :P [18:43] s/pratitions/partitions/ [18:43] Yes, either works [18:43] RoyK: Should I delete the partition ? [18:43] Schalla: just create the raid [18:43] Schalla: on the drives directly [18:44] Schalla: it'll overwrite those partition tables nicely [18:44] Alright :) [18:44] as long as you don't have to worry about booting on them [18:44] patdk-wk: he has a couple of USB sticks for the root [18:44] fine then [18:45] Yeah just as iso / kvm disk storage [18:45] I use partitions so that tools other than the linux kernel knows what they are, but whether that matters is up to you [18:46] Uhm. [18:46] I am confused. [18:46] Sec, paste incoming [18:46] qman__: I think you need to be particularly interested to do that ;) [18:46] Confusion cleared, forgot that I created 2 raid devices for swap raid and boot raid [18:47] s/boot/root [18:47] :) [18:47] Everything worked out great RoyK btw! [18:47] nice [18:47] * RoyK Likes Things That Works (tm) [18:47] :> [18:48] It can also matter more if people who don't understand it work on your systems [18:48] Schalla: did you get another usb stick too? [18:48] RoyK: Ordered [18:48] Didn't receive it yet [18:49] qman__: it's a problem if you let people without good knowledge of Linux have root access to your system, yes, but that's a completely other issue [18:49] qman__: Yeah thats the reason I want to learn to destroy my own systems before I do anything on other peoples systems. :p [18:50] :D [18:50] that's the spirit [18:50] I unfortunately have to deal with that at work, where only about 4 out of the 50 or so techs know linux [18:50] RoyK: The USB stick I was using is out of stuck and can only be delivered at the 19 january, however, guess its fine to run with a single stick till then? [18:51] qman__: you give those guys root access? [18:51] (Not like I got another option.. :D) [18:51] Everyone has it, not my option [18:52] I guess its fun when multiple people work with root at the same time on the same host. [18:52] qman__: ouch [18:53] So I try to make everything I do as intuitive and easy as possible, and write detailed documentation, etc. [18:53] RoyK: after md2 was created, start "parted /dev/md2, create a new partition table with "mktable" and then just create one partition with "mkpart primary ext4"? [18:54] Schalla: no... [18:54] Schalla: don't use partitions on a raid [18:54] :| [18:54] Schalla: vgcreate mysuperspecialdata /dev/md2 [18:54] Schalla: lvcreate -L 1T -n littleterabytevolume mysuperspecialdata [18:54] just wondered because the other raids do so? [18:54] for instance [18:54] Schalla: just use lvm, ok? [18:55] Use lvm on raid, not partitions [18:55] Yeah makes sense to me [18:55] Just wondering, why does Ubuntu doesn't do so by default? Is there a easy answer to it? (Talking about the software raid wizard) [18:55] Schalla: and if you want to grow that filesysstem to > 16TiB some day, use xfs, not ext4 on top [18:56] Schalla: iirc there isn't a wizard, just a way to create raid things and raid on top. after that you can create lvm things on top of the raid [18:56] I always do that [18:56] Guess I simply skipped that [18:56] https://help.ubuntu.com/community/Installation/SoftwareRAID#Partitioning_the_disk [18:57] ^ followed that guide [18:57] Which is somewhat outdated [18:57] Anyway, will follow the advice and use lvm, thanks. :) [18:57] Schalla: doesn't say anything about lvm [18:57] Schalla: but it doesn't hurt not using it, but on a raid like that, I *always* do [18:58] To expand the storage pool later on easily? [18:59] Schalla: well, to make everything more flexible [18:59] Schalla: better start out with a filesystem of 1TB or so and expand later [19:00] Alright [19:00] Schalla: you may want to separate the iso files and the kvm images - no idea - but you may - and so on [19:00] Schalla: so start out small (100GB?) and expand later - you won't even need a reboot - no downtime [19:01] with ext4 you may even shrink them [19:01] It makes sense I guess. Should I go for both ext4? I dont plan to get more than 8tb storage [19:01] that's not possible with xfs, though [19:02] ext4 works well up to 16TB although it's a PITA when it comes to fscking a bad volume (you'll probably have to run one at some time) [19:02] or every 180days :) [19:02] patdk-wk: tune2fs -c0 -i0 [19:02] Just installed lvm2, got a warning of mdadm that the newly created raid is not in mdadm.conf yet, thats normal? [19:03] Schalla: mdadm --detail --scan >> /etc/mdadm/mdadm.conf [19:03] Schalla: and run update-initramfs -u [19:04] http://hastebin.com/iciyisupid.sm [19:04] Thats normal that the generated ones are /dev/md/[01] and mine is /dev/md2? [19:04] looks fine - perhaps change md/[01] to md[01] in that file (just edit it) [19:05] or just don't worry [19:05] it's the same thing [19:05] Schalla: did you use > /etc/mdadm/mdadm.conf or >> /etc/mdadm/mdadm.conf ? [19:05] >> [19:05] Just didnt copy it entirely [19:06] good [19:06] I appended didnt overwrite [19:06] [A [19:06] good [19:06] No errors now. :) [19:07] ok, create an lv and put a fs on it [19:08] vgcreate / lvcreate / mkfs -t ext4 (for instance) [19:08] Do I run the mkfs -t ext4 then on /dev//? [19:08] mhm [19:10] then perhaps tune2fs -c0 -i0 to stop it fscking the filesystem regularly (you don't need that) and perhaps tune2fs -m.1 or so to reduce the normal 5% reserved space to root to 0.1% [19:10] Worked like a charm. [19:11] then make a mountpoint, edit /etc/fstab with appropriate data and run 'mount -a' [19:13] I would use the LV UUID to create the mountpoint or rather stick witht LV path? I think to remember for disks uuid was recommended, but unsure aboutlv [19:13] *lv [19:13] the lv path [19:16] And once again, everything working fine. auto mount works, file creation works, filesize is correct, software raid is stable. :) [19:18] :) [19:18] Schalla: give it a reboot to see if it comes up correctly [19:18] yup, good idea [19:20] the reason I thought of to put kvm image separately, was that they like to live in /var/lib/libvirt/images and you may not want the iso files around there, so two separate mountpoints may be a good idea (up to you of course) [19:24] Schalla: some use bind mounts, but I've learned over time that those tend to break at updates - good old mounting things at the right place works better [19:24] RoyK: Nope, doesn't come up. But might be related to the network, I sadly gtg once again :| I will check it when I am back later tonight [19:24] Thanks again for everything [19:25] (Just tried whether I can SSH to it, that didn't work. Unsure what this is related to) === Lcawte|Away is now known as Lcawte [20:14] coreycb: the only one blocker to bump our CI to mitaka is having aodh working on trusty [20:16] EmilienM, ok. do we just need aodh-api to pull in pymongo? if so that's just awaiting upload. [20:17] coreycb: http://logs.openstack.org/35/255635/16/check/gate-puppet-openstack-integration-scenario001-tempest-dsvm-trusty/f4ab95b/console.html#_2016-01-12_16_22_42_991 [20:18] EmilienM, have you checked with upstream on that? I'm not sure if that's a packaging issue or not. [20:18] coreycb: well, it works on master with RDO [20:19] the same job, same config [20:19] EmilienM, ok [20:19] NoSectionError: No section: 'alembic' [20:19] EmilienM, I'll dig into it when I get a chance [20:21] coreycb: I'll try to find the issue [20:21] do you test aodh? [20:22] EmilienM, we've just been syncing with debian thus far but are carrying a small delta just for some dependency adjustments atm [20:23] coreycb: small delta from debian, right? [20:23] EmilienM, and we've not finished moving the ceilometer charm to using aodh yet [20:23] EmilienM, yes [20:23] ok === jamietech is now known as Guest13549 [21:00] Figured it out guys [21:00] just for the record.... it turned out to be a Biosdevname issue. [21:02] the jetpack i was using presents itself as just another ethernet connection. It then sets up the routing side for the modem...dont need wvdial or anything else really. [21:04] as soon as i disabled Biosdevname and renamed my interfaces accordingly, added interface for eth4 and set to dhcp...(I used udevadm monitor to find out what eth device was being assigned). I rebooted and it came right up and started working. [21:18] grendal_prime: Nice :) [21:21] now...i dont have a verizon signal here.. to get all the way to the internet with...but, i can ping the gateway on the device itslef..so it should be fine. [21:22] im using these little guy to provided my required "redundant network" connectivity for several remote pieces of equipment. === Lcawte is now known as Lcawte|Away [22:44] Back again, problem solved that my server didn't boot, mdadm did something weird. [22:47] For some weird reason when I created /dev/md2 he created the file as "md/" in the folder "/dev" [22:47] I can't explain yet how that could happen [22:47] Command used to create the raid: "mdadm --create --level=1 --raid-devices=2 /dev/md2 /dev/sda /dev/sdb" [22:48] look in /etc/mdadm/mdadm.conf [22:49] the path it uses is in there [22:53] rattking: I did, and the weird thing is that my 2 first created arrays were generated as /dev/md/0 and /dev/md/1 while the new one as /dev/md2 [22:54] I had to manually restore the old names and comment out the md2, since that resulted in a boot error [22:54] are all 3 arrays in that file then? [22:56] No, since I had to delete the newly created array that the system booted at all [22:57] I just deleted the entry, booted the system removed the junk file, md0 and md1 and still working as before, md2 disappeared [22:57] I executed the same command again, md2 is now shown again as active + syncing [22:58] Hm, the bug didnt show up either [22:58] :| [22:59] bug? [22:59] Well, error [22:59] Bug is a bad word, I just am unsure what I did wrong. [23:18] Or well, I thought I fixed it. Created the raid again, created the vg, lv, fs and mounted it, created a file, everything working fine. [23:18] ./dev/md2 existed, mdadm showed its status correctly [23:18] I executed update-initramfs -u and restarted [23:19] => /dev/md2 is gone, boot again failed [23:25] I have having trouble getting an mysql server to run [23:26] it was working fine yesterday but then i did a package update and now it fails to start [23:26] the error i am getting is [ERROR] [redacted]/mysql/bin/mysqld.bin: unknown variable 'defaults-file=[redacted]/mysql/my.cnf' === logan2 is now known as logan-