[00:00] blizzow: do you have some other package installed that maybe is "watching" eni? [00:04] It's a fresh install with the only thing selected during the install being openssh server. To be fair, this is happening via ansible. I copy /etc/network/interfaces containging a static ip into place, and the next command ansible tries to run fails because it can no longer connect to the host. [00:06] blizzow: i'm not sure if ansible would be interacting there. I woudl maybe try it without ansible (seems like it would be trivial to do) [00:07] blizzow: that is, fresh install, ssh in, change eni (or more appropriately, maybe eni.cfg/... [00:15] I copied a new /etc/network/interfaces over manually and it won't change over now. Like I said, it's 50/50 for me so far and I can't figure out rhyme or reason as to how ubuntu is dealing with the change of /etc/network/interfaces. [00:16] blizzow: well, like I said, it's rather unexpected to me (and I've never seen that), so I'm not convinced it's ubuntu doing anything yet. AFAIK, various tools might write to eni (and ifup/ifdown read from it), but it doesn't happen spontaneously === arooni is now known as arooni__ [02:12] Hi good ubuntu folks. I attempted to upgrade my server from 14.04 to 16.04 tonight and the installation failed. When I try to get things working again with the "apt-get -f dist-upgrade" command to the packages it fails on "util-linux" package with "insserv: Service mountdevsubfs has to be enabled to start service hwclock" ... while I have a backup, I figured I'd play around with this botched system as a learning experienc [02:21] TylerWhitney: odd indeed. could you file a bug against util-linux and initscripts about that? [02:22] TylerWhitney: for your own playtime this may be helpful : find /etc -name '*hwclock*' and find /etc -name '*mountdevsubfs*' [02:24] I will file a bug, in my original searches I did see a similar mention of this, but I'll have to go back to google to make sure it was filed as a bug [02:24] sarnold: With those find commands I see init scripts for both of them [02:25] If I try to run the script for mountdevsubfs I get the message "Can't open /lib/init/tmpfs.sh" [02:25] TylerWhitney: the S symlinks say when to start the service during the boot; the K symlinks say when to shut down the service when changing runlevels or shutting down [02:26] strange, I don't have that file either [02:27] appears to be called on line 26 of the init.d file [02:30] I created a quick and dirty tmpfs.sh script there that creates a loopback on a tmp file and mounts it [02:31] when I run the init script it says mountdevsubfs is running, though I get the message still when trying "apt-get -f install" that it hit a problem with util-linux because "Service mountdevsubfs has to be enabled to start service hwclock" [02:31] very odd [02:35] now try the 'update-rc.d mountdevsubfs enable' -- maybe with a -n to prevent it from doing anything, so you ca nsee what it intends to do first [02:37] hmm "update-rc.d: error: cannot find a LSB script for mountdevsubfs" [02:42] try with mountdevsubfs.sh ? [02:43] Yes, that was the issue I think. [02:44] I also tried removing it with the update-rc.d command and then setting it with defaults [02:44] That seemed to get traction [02:44] I still got an error with "apt-get -f install" but the command "apt-get -f dist-upgrade" is working and installing packages now... [02:44] This seems promising. [02:45] But you were right I forgot the .sh at the end when I first ran it, hence the not found error [02:45] I think I also know what caused the issue in the first place now. [02:47] oh? :) [02:47] So, I had a really old custom written init script in the init.d directory [02:48] I think because it didn't conform to standards when it came to updating it somehow failed starting/restarting all the services [02:48] I'm more of an application developer that dabbles in this stuff, usually when things go awry I just restore from backups... I'm a lazy sys admin [02:48] But figured tonight was a good time to play on a broken system... you never learn as well as you do when fixing broken stuff [02:48] I [02:49] I'd hope it would take more than that to send it awry, but maybe that was it. :) [02:49] Yes, I could be very wrong [02:49] I saw an error relating to trying to start that old service that no longer existed when I ran "dpkg --configure -a" and thought... oh damn that doesn't exist anymore [02:50] well I made some progress and now we fail on mysql-server [02:56] Seems to be a known bug http://askubuntu.com/questions/760724/16-04-upgrade-broke-mysql-server [02:56] I did an "apt-get remove --purge mysql-server mysql-server-5.7" then deleted everything in the /etc/mysql directory and did an "apt-get install mysql-server mysql-server-5.7" and it worked like a charm [02:57] And... thats it, we're back to normal and 16.04 it appears! [02:58] sarnold: thanks for the push in the right direction [02:58] I was not familiar with all of the update-rc.d commands [02:58] hopefully we'll be rid of update-rc.d some day [02:58] He [02:58] Well, that was certainly the issue... the MySQL server thing appears totally isolated and separate from the first [02:58] yeah it probably is [02:59] this was a production server, I like living on the edge! [02:59] Its a minor server and I schedule the downtime for tonight though ;-) [02:59] ahhh at least it happened on your terms. sortof. :) [03:01] Haha! Now the real deal, lets see what happens when I reboot [03:01] I have the backup image ready if need be. [03:03] The right way would probably be to install 16.04 fresh, install the required packages, then restore files from backup. [03:03] But then, what fun would that be. [03:04] the mysql issue seems annoying. Doing the purge approach to fix it seems needlessly brutal .. [03:04] 'cause now you've got to deal with the mysql configuration from scratch. [03:13] luckily my configuration wasn't all that customized [03:13] I seem to have run into more issues with PHP considering it seems php7 is now default. [03:14] Got everything working again, but had to manually install some packages that were there before trying to upgrade [03:14] s/default/only php/ [04:20] /msg NickServ VERIFY REGISTER TylerWhitney oqvrpwipphlb [04:20] fail [06:19] when the OOM-killer is invoked, is it possible to execute a script? Since I know for 99% which services would need to be restarted. [06:22] ndee: You can probably hack something together by have something watch the kernel log, but I don't think there is any explicit reliable way to hook a script in there. [06:23] andol: ok, since normally, it's always mysql and apache that needs to be restarted and the server would function again but I'm not sure on how to achieve that. [06:23] ndee: Depending on circumstances, I'm otherwise partial to have the kernel panic-and-reboot in case of an OOM. [06:25] Yet, in case this is a common problem for you, it sounds like the first thing you should do is to tune the memory usage of Apache and Mysql. [06:27] hi all i have this problem after upgrade from ubuntu 12.04 ==>14.04 [06:27] dpkg: dependency problems prevent configuration of libapache2-mod-php5: [06:27] libapache2-mod-php5 depends on apache2 (>= 2.4); however: [06:29] andol: I just checked, in apache, the processes just started to rise and rise until 256 and then remained there until the restart of apache. How could I avoid that? [06:31] the keepalivetimeout is set to 5 and the timeout in general is set to 300. The number of processes rose over a time of 6 hours [06:33] ndee: Was a while since I spent some serious quality time with Apache, but if I remember correctly you might want to tune your MPM settings, regarding how many childs and threads to allow, etc [06:35] ndee: But really, now I'm mostly shooting from my hip, so you should probably rather get your advice from someone who does this stuff for real. === terje is now known as Guest20636 === _degorenko|afk is now known as degorenko === iberezovskiy|off is now known as iberezovskiy [10:35] will 14.04 systems be getting automatic upgrade soon? [10:37] FManTropyx: Yep, for some value of "soon". :) [10:38] okay, I remember hearing today's date previously [11:01] Isn't it safe to uninstall mdadm and lvm2, when its not needed? I have ubuntu-server installed, and these two packages are depended upon [11:01] I find the message "W: mdadm: /etc/mdadm/mdadm.conf defines no arrays." a real annoyance [11:04] must have picked the install everything option, instead l minimal install [11:04] I actually find having openiscsi installed to be more annoying [11:06] patdk-lap: I haven't noticed that one actually. How is it bothersome? [13:21] jcastro: hey, do you know if there's a PHP webapp charm? I need to set up a LAMP stack for a non-work project but I couldn't find one. [13:21] ddellav, coreycb: ceilometer and aodh are now blocked on cotyledon [13:21] well for post b2 anyway [13:50] Hi, I was trying to assign group permissions in my server, can someone help me out setting up group permissions on the system level rather than file/directory level? [14:13] rbasak: hey, I'm not sure yet if I find the time - but atm I'm kind of blocked on almost all ends I worked on, so would you pick 1-3 bugs depending on their compexity for me to work on? [14:17] cpaelzer: I don't have many prepared, sorry. I need to fill the backlog again. How about bug 1594202? Or just take a look around for one? [14:17] bug 1594202 in dovecot (Ubuntu) "apparmor messages everywhere" [Medium,Confirmed] https://launchpad.net/bugs/1594202 [14:18] cpaelzer, or knock through some more merges too [14:18] cpaelzer: see "List of recently modified bugs in packages looked after by the server team" in https://wiki.ubuntu.com/ServerTeam/KnowledgeBase [14:21] jgrimm: do you still have the generated list for the merges that could identify candidates other than just looking around? [14:21] jamespage, ok and as for MIRs I think we're just blocked on aodhclient and ryu [14:23] cpaelzer, i have a script now! but really its easy enough to just look at MOM [14:23] sure, can do that in case I run dry again :-) [14:23] cpaelzer, maybe rbasak would hand off exim4 to you? i know that's been on his backlog for a long while? [14:23] cpaelzer, and i'm quite certain rharper would let you take strongswan [14:24] I didn't want to grab something too big before vacation as surely curtin, dpdk and the open merges are soon suddenly full of work once unblocked [14:24] fair enough, plenty of those too! [14:27] cpaelzer, logwatch and amavisd-new are probably trivial [14:27] rbasak: nacc: so to be ready for safety if one could make available for the importer: exim4, logwatch and amavisd-new - that would be great no matte rwho takes it eventually [14:28] jamespage, ddellav: zigo's working on cotyledon. we'll need to MIR that as well. [14:28] cpaelzer: sure, I'll import now. [14:28] coreycb ah ok, good news [14:29] coreycb, ack [14:29] rbasak, 'at' and memcached too please [14:30] ack [14:33] jgrimm: https://bugs.launchpad.net/usd-importer/+bug/1605280 [14:33] Launchpad bug 1605280 in usd-importer "memcached cannot be imported" [Undecided,New] [14:34] rbasak, :) thanks [14:38] jamespage, ddellav: aodhclient MIR approved [14:42] cpaelzer: if you are looking for a bug to solve, https://bugs.launchpad.net/ubuntu/+source/percona-xtradb-cluster-5.6/+bug/1574509 would still be a good one for me ;) [14:42] Launchpad bug 1574509 in percona-xtradb-cluster-5.6 (Ubuntu) "mysqladmin-5.7 fails to set password for percona-xtradb-cluster-server-5.6" [Undecided,Confirmed] === jelly-home is now known as jelly [15:50] rbasak, fyi->https://bugs.launchpad.net/usd-importer/+bug/1605280 [15:50] Launchpad bug 1605280 in usd-importer "memcached cannot be imported" [Undecided,New] [15:51] err meant nacc ^^ sorry [15:52] jgrimm: yep, i'm reading and working on it now [15:52] nacc, cool. tx [15:52] how do I stop a tty instance and disable it in Ubuntu as is done here for CentOS: https://mrkmg.com/posts/2015/07/proxmox-centos-7-console-showing-two-login-prompts/ [15:55] coreycb lp:~ddellav/ubuntu/+source/ceilometer and lp:~ddellav/ubuntu/+source/openstack-trove ready for review/push. FYI: Trove had quite a few dependency updates. I'm not sure what you want to do about that but i added them. [16:00] ddellav, coreycb: newton-proposed passed tempest smoke - pushing to updates now [16:00] \o/ [16:01] woot [16:09] ddellav, ceilomter is pushed/uploaded [16:13] ddellav, for openstack-trove the only dep that should need updating is oslo.concurrency>=3.7.1, which is ok. [16:13] coreycb there was a bunch that were out of sync. Do we just ignore those? [16:14] ddellav, that's the only one I noticed was different between 5.0.0 and 5.1.0 [16:15] coreycb thats correct but if you manually diff the requirements.txt and d/control there are many differences [16:15] coreycb https://cl.ly/2s2u331N1D0A/Image%202016-07-21%20at%2012.15.28%20PM.png [16:17] ddellav, gotcha, so we missed updates to d/control on previous changes. /me looks again. as long as they're in the archive for xenial it should be fine. [16:19] ddellav, yep looks good [16:20] coreycb ok, then it's good to push :) [16:20] ddellav, yep, builds ok I assume? [16:20] coreycb yessir [16:25] ddellav, trove pushed/uploaded as well. for ceilometer I asked arges to reject our previous mitaka point release that was in the review queue since this one overrides it. [16:26] coreycb ok, i updated the bug and the card to include 6.1.3 as well [16:42] hey guys is there a reason why 16.04 doesnt require sudo when one does init 0 or what ever init command [16:44] eagles0513875|2: yes [16:45] tarpman: whats the reason is it expected behaviour or a bug [16:46] eagles0513875|2: the default policy allows users with a local session to shut down the system. it's expected. [16:47] isnt that a bit of a risk especially for production systems [16:47] and when you say policy what do you mean exactly by policy please? [16:50] coreycb: which one [16:50] 6.1.1 [16:50] arges, yes 6.1.1 [16:50] thanks [16:50] coreycb: ok done [16:52] eagles0513875|2: /usr/share/polkit-1/actions/org.freedesktop.login1.policy is the policy I'm talking about. specifically with yes [16:53] ok never really looked into that stuff that is why it works and i dont touch but thanks for enlightening me :) === iberezovskiy is now known as iberezovskiy|off === degorenko is now known as _degorenko|afk [18:07] hey guys do you know how to clone a disk? http://unix.stackexchange.com/questions/297446/how-to-clone-an-entire-linux-disk-and-boot-it [18:10] poood: dd? [18:10] dr4c4n: i used dd but forget the boot sector [18:10] don't know how... [18:14] it shows you on the page how to backup the mbr [18:14] if you clone the whole disk you don't need to worry about a disk sector specifically; if you clone just a partition then you'll have to install a boatlaoder of some sort [18:14] sigh [18:15] yes then run grub-install [18:16] sarnold: okay. thank you. let me try :) [18:16] many thanks btw [18:16] so helpful [20:20] hi, im trying to upgrade from 10.04 to 12.04 with do.-release upgrade [20:21] if finds precise 12.04 but then stops after reading repos and restores back to previous source list [20:21] any idea why is this? [20:21] I could paste the output but since its long I dont know how to scroll as the process from what I can understand is run inside the "screen" utility [20:22] hi, I'd like to download the xenial cloud image (which I've done) and add a bunch of packages to the img file. [20:22] I've attempted to guestmount it, then apt install [20:22] but I'm getting this error: [20:22] Fatal Python error: Failed to open /dev/urandom [20:23] any suggstions? [20:24] https://ubuntuforums.org/archive/index.php/t-2305183.html [20:26] anyone? [20:27] terje: are you mounting and chrooting into a image? if so you probably need to bind mount proc sys and dev into the images filesystem.. [21:01] I found the isse in the logs, at a certain point in the do.release upgrade at the beginning it says something like [21:02] "it was not possible to authenticate some packages. This could be a temp problem etc" [21:02] any clue? [21:03] I *think* the package signing keys are in ubuntu-keyring maybe you need to update that first? [21:04] I did that [21:04] a lot of times [21:04] I did an apt-get update and upgrade [21:04] rebooted [21:04] I did them all [21:04] now I was reading this http://askubuntu.com/questions/425355/error-authenticating-some-packages-while-upgrade [21:05] did you pull in the package from the version you are trying to upgrade to? [21:05] its a long story, im on 10.04 [21:06] I put the EOL repos [21:06] they seem to work [21:06] but do-release upgrade is failing [21:06] miserably with that log (and it took me 1 hour to find the log!" [21:06] but at least now I know whats happening [21:25] yea, http://askubuntu.com/questions/425355/error-authenticating-some-packages-while-upgrade fixed my problem === arooni__ is now known as arooni [21:45] Hi guys, I have problem with kvm/libvirt and apparmor. Usb-passthrough doesn't work because of apparmor profile. Question is that I can't find answer anyware, are quotes considered to be ok in apparmor profile defined by libvirt-xxx.files? === JanC is now known as Guest9881 === JanC_ is now known as JanC [21:48] vidasov: yes, you can quote paths with e.g. spaces in them if you need to [21:49] ok, sounds good but that is not a bug which bothers me than. thnx sarnold [21:54] but why some paths in libirt-xxx.file are quoted and some not, no spaces in path. I gues if quotes are ok than I should not consider it anymore.... [21:56] vidasov: the paths may have been stuck in quotes by an automated tool somewhere, which didn't go to the trouble of finding out if the quotes are needed :) [21:57] Yes, looks like I need to oppen a bug request :-(