[00:00] conrmahr: dd if=/dev/zero of=/dev/thatdevice bs=1M count=1 [00:02] that removed the RAID array? [00:03] now i want to format with zfs [00:03] can anyone help me authenticate to my samba server via ldap on a windows server please ? [00:04] first mdadm --stop the raid [00:04] then mdadm --zero-superblock those devices [00:04] then zpool create yourpool ..... [00:04] Did that, but it says Cannot get exclusive access to /dev/md2 [00:04] do you have a vg on that? [00:05] yes [00:05] then vgremove [00:05] pvremove [00:05] etc [00:05] i don't know what it means [00:05] but i have it [00:05] in the fdisk -l [00:05] pastebin "pvs;lvs;vgs" [00:07] do i have to install a pkg for pastebin? [00:07] !pastebinit [00:07] oh wait i'm stupid [00:07] pastebinit is the command-line equivalent of !pastebin - Command output, or other text can be redirected to pastebinit, which then reports an URL containing the output - To use pastebinit, install the « pastebinit » package from a package manager - Simple usage: command | pastebinit -b http://paste.ubuntu.com [00:09] http://paste.ubuntu.com/16450099/ [00:10] you haven't stopped the raid, have you? [00:10] cat /proc/mdstat [00:11] no [00:11] it says i don't have special access [00:11] make sure you use mdadm --zero-superblock on those disks before you create a zpool [00:11] well, vgremove, pvremove [00:12] and perhaps umount the filesystem first [00:12] i don't know the cmd line [00:13] sudo vgremove /dev/md2 [00:13] ? [00:13] pastebin output of "mount" and "lvs" [00:14] http://paste.ubuntu.com/16450137/ [00:15] sudo lvs = no volume groups found [00:15] well, you have a zpool [00:16] pastebin lsblk output [00:16] and zpool status [00:17] just fyi, i have 1 SSD 14.04 UBNT disk, and 2 WD Red 4GB (just for data disk) [00:17] second one i'm trying to clean [00:17] please do as I asked [00:18] http://paste.ubuntu.com/16450202/ [00:18] and yes I am an idiot [00:19] mdadm --stop /dev/md2 [00:19] mdadm --zero-superblock /dev/sdc5 [00:20] zpool attach --force data1 sdb1 sdc1 [00:20] or something like that [00:20] mdadm: Cannot get exclusive access to /dev/md2:Perhaps a running process, mounted filesystem or active volume group? [00:20] you may want to dd a few zeros over that disk first [00:20] what about vgs? [00:20] pvs? [00:20] lvs? [00:21] vgs first [00:21] sudo vgs [00:21] ? [00:21] yes [00:21] No groups found [00:21] pvs? [00:22] nothing lists [00:22] lvs no groups found [00:22] dd a bunch of zeros over sdc [00:22] what's strange is that lsblk lists is as a member of lvm [00:22] well this drive [00:23] i took out of a Synology Diskstation [00:23] well, just dd over it [00:23] so what now [00:23] sudo dd if=/dev/zero of=/dev/md2 bs=1M count=1 [00:23] no [00:24] the disk, not md2 [00:24] md2 was the raid [00:24] ok [00:24] just don't overwrite sda or something :P [00:24] what do you want to do [00:25] dmraid -r -E /dev/sdc if you want to remove raid flag [00:25] doesn't matter [00:25] dmraid? [00:25] that's OT in here [00:26] sudo dd if=/dev/zero of=/dev/ sdc=1M count=1 ? [00:26] that's what i had to do a few times to get raid disk recognized :p [00:26] conrmahr: to sdc, probably [00:27] conrmahr: that wipes the first 1MB of the disk [00:27] ok i did [00:27] conrmahr: I won't post the full commandline - that's not accepted [00:27] conrmahr: ok - try lsblk again [00:28] not accepted? lol [00:28] sudo dd if=/dev/zero of=/dev/sdc bs=1M count=1 [00:28] bindi: just because someone may do that unintentionally [00:28] :| [00:28] look the same [00:28] conrmahr: give it a reboot [00:29] shutdown -r [00:29] or just "reboot" :P [00:29] oh i forgot simple reboot [00:29] init 6 [00:29] :P [00:29] :) [00:30] ok looks like it's removed! [00:30] the "reboot" command has been around for almost 10 years ;) [00:30] sdc 8:32 0 3.7T 0 disk [00:30] only [00:30] ok - so do you want those mirrored? [00:30] yeah it's just a backup [00:31] zpool attach data1 sdb sdc [00:31] should work [00:31] ugh [00:31] perhaps with an -f [00:31] creating a new pool? [00:31] i used zfs to format the first one [00:31] just attach the other disk [00:32] it'll become a mirror [00:32] zpool create -f -m /mnt/mypool mypool mirror ata-1234 ata-2345 [00:32] ? :P [00:32] ls /dev/disk/by-id [00:32] you really should do it by-id [00:32] doesn't matter [00:32] sure it does [00:32] no. it. does. not. [00:32] if you use a script to mount them then it does :p [00:32] zfs will revert to using by-id names [00:33] nvalid vdev specification [00:33] use '-f' to override the following errors: [00:33] /dev/sdc contains a corrupt primary EFI label. [00:33] conrmahr: as I said, you may need -f [00:33] i think i need to format it first right? [00:33] no [00:33] just use -f [00:34] ok [00:34] that will create the EFI label [00:34] there's nothing like formatting anymore [00:34] look like it did something [00:34] how do i name the drive? [00:34] conrmahr: try zpool status [00:34] or mount it [00:34] conrmahr: it's mounted under /data1 [00:35] conrmahr: perhaps name that "data" [00:35] just export it and reimport it as "data" [00:35] http://paste.ubuntu.com/16450521/ [00:35] you'll have subvolumes for that [00:35] goodie [00:35] looks good [00:36] beautiful [00:36] so i don't need to name it /data2 [00:36] if you want to rename it, export the pool and import it with a new name [00:36] Is there really no way to get PHP5 working on Ubuntu 16.04? [00:37] There are a lot of things that I can't get to work on it right now, because they all want PHP5 [00:37] if i new how to do that [00:37] knew [00:37] conrmahr: zpool export data1 [00:37] Sachiru: http://askubuntu.com/questions/756879/cant-install-php5-on-ubuntu-16-04 [00:37] conrmahr: zpool import data1 mynewname [00:37] check out the ppa [00:39] RoyK: so this will name the first drive /data1 and the second /data2? [00:41] Sigh, thanks. So much is broken with PHP7 (omdistro, librenms, etc.) [00:42] But I guess that isn't ubuntu's fault, rather, they don't move to PHP7 fast enough. [00:42] conrmahr: no - the two drives are mirrored [00:42] conrmahr: meaning when one of then dies, the data persists [00:42] so i'll just keep it as data1 [00:42] no need to change [00:43] Ooh, we have ZFS discussion here? [00:43] Nice! [00:43] conrmahr: if you want to stripe them, which I won't recommend, just detach sdc and add it [00:43] but don't do that, really [00:43] Sachiru: Not really I used it to setup my first drive [00:44] yeah I like to mirror, that's all I wanted anyway [00:44] is it clustering the data? [00:44] drives die, the silicon wants to back to the mountains [00:44] it's all natural [00:44] conrmahr: Define "clustering" [00:45] Because it means many different things, some of which apply to zfs, some of which do not. [00:45] conrmahr: it's not clustering, that's far larger [00:45] file gets written to Disk1, then soon after it its copied to Disk2 [00:46] conrmahr: no, they are written to both at the same time [00:46] i only know clustering from the work i've done on MariaDB between to server databases [00:46] conrmahr: and when one drive files (not "if"), the data is available [00:46] ok even better [00:47] how would you know if one drive fails/ [00:47] s/files/fails/ [00:47] zpool status shows that [00:47] or use some monitoring [00:47] smartmontools is good at that [00:47] do i have to apt-get that? [00:48] install smartmontools and it installs smartd which can send you emails when something goes wrong [00:48] yes [00:48] awesome [00:49] ZFS can detect if a disk fails (as in whole disk fails), or a certain sector/cluster on the disk fails [00:49] If the cluster fails, ZFS can detect if it can repair it or not, and auto-repairs if it is repairable [00:49] If not, it reports an unrepairable failure via "zpool status " [00:49] Sachiru: sometimes smartmontools is good also, to detect pre-failures [00:50] You can also initiate a scrub ("sudo zpool scrub "). This reads all data written to disk, checks against checksum, and repairs repairable errors. [00:50] this is great [00:50] smart stuff doesn't always work as intended, but according to google disk stats, almost 50% of the failures gave smart errors before dying [00:51] failures as in smart errors [00:51] Additionally, (if you have Ubuntu installed onto a ZFS dataset as root), you can snapshot and revert easily. Thus, you can do "apt-get dist-upgrade" and other potentially destructive features without fear, since you can just snapshot the dataset, and reboot to the snapshot if it fails. [00:52] Same thing with if you want to install anything. Also, snapshots take at most 5 seconds, even if the thing you're snapshotting is several hundreds of terabytes of data. [00:52] Reversion takes the same amount of time. [00:52] so the config file says don't use SMART if you are using smartd [00:52] * RoyK messed up a VM rather badly last night, but then, it was on ZFS, so he just restored from the 15min old snapshot - zfs autosnap is neat [00:53] conrmahr: huh? [00:53] conrmahr: it's the same thing [00:54] # List of devices you want to explicitly enable S.M.A.R.T. for [00:54] # Not needed (and not recommended) if the device is monitored by smartd [00:54] conrmahr: btw, if you have drives supporting SCTERC, turn that on if it's not enabled already [00:54] conrmahr: no need to list any [00:54] ok [00:54] I have WD Red 4TB NAS (2x) [00:54] do they support SCTERC? [00:55] conrmahr: can you pastebin smartctl -x /dev/sdb? [00:56] RoyK: Why the conservative auto-snapshot? [00:56] Sachiru: conservative? [00:56] http://paste.ubuntu.com/16450748/ [00:56] I have mine snapshot every 5 minutes. Then again, the host that does this handles only 6 VMs, and most of them are nginx+php+mariadb stacks. [00:57] Sachiru: see SCT Error Recovery Control in there - set to 7 seconds [00:57] should work fine [00:57] I prefer 1s or so, but that's up to you [00:57] RoyK: I mean zfs auto-snapshot [00:58] Not SCT [00:58] For VMs [00:58] conrmahr: sorry, that was for you [00:58] Ah [00:58] Sachiru: I beleive 15 minutes is sufficient [00:58] so in the smartd.conf? [00:59] not sure - but anyway - 7s is ok [01:00] far better than without ERC [01:00] dont you want to disable that tler thingy? [01:00] without ERC the disk can go into so-called deep recovery, meaning it'll spend a minute or two trying to recover a single sector [01:01] bindi: ERC == TLER - and that's not a thing you want to disable [01:02] without ERC, your raid, be it md or zfs, may kick out a drive for a single bad sector [01:03] RoyK: I think by default it's set to 7s [01:03] i did $smartctl -l scterc /dev/sdc [01:03] Read: 70 (7.0 seconds) [01:03] Write: 70 (7.0 seconds) [01:03] 7 is default on sata disks [01:04] so I should change to 1s? [01:04] whats the advantages and disadvantages? This is a NAS/Media Server [01:05] advantages are to avoid a 7s drop if a sector goes bad [01:05] zfs can handle that [01:06] disadvantages are (something Donald Trump said) [01:07] IMHO, it's good to set ERC even if you're not using ZFS [01:07] conrmahr: smartctl -l scterc,10,10 /dev/something [01:08] I know mdadm (at the very least) also complains about dropped sectors. [01:08] Sachiru: it's not about sectors [01:08] Sachiru: it's about the disk trying to find out about those sectors and becoming unavailable for a long time [01:08] RoyK: I know, it's sectors causing drives to drop from the array [01:09] and you don't really want a few dead sectors to make your md or zfs or whatever to kick it out [01:10] that's why we have raid [01:10] Write SCT (Set) Error Recovery Control Command failed: scsi error badly formed scsi parameters [01:10] SCT (Set) Error Recovery Control command failed [01:10] Retry with: 'scterc,70,70' to enable ERC or 'scterc,0,0' to disable [01:10] Write SCT (Set) Error Recovery Control Command failed: scsi error badly formed scsi parameters [01:10] SCT (Set) Error Recovery Control command failed [01:10] Retry with: 'scterc,70,70' to enable ERC or 'scterc,0,0' to disable [01:11] heh - crippled fucking disks [01:11] I've stopped bying WD stuff [01:11] but then, 7s should do [01:11] what the? [01:11] WD cripples the firmware [01:12] toshiba has good, cheap drives with good firmware, at least for now [01:13] the 'enterprise' SATA disks from WD are just the same as the desktop drives, just with better firmware [01:13] a few years ago, they were the same, more or less [01:14] how do i start smartd [01:15] which ubuntu version_ [01:15] ? [01:15] trusty [01:15] should be running [01:15] service smartd start [01:15] you didn't change the config file? [01:16] how do i define my email for notifications? [01:16] default config should work [01:16] it sends email to root [01:16] just forward root emails to yourself [01:16] ah right [01:16] in /etc/aliases [01:16] (and then run newaliases) [01:18] no such dir [01:20] is there any mail in the queue? [01:20] mailq should tell [01:21] do you have postfix installed? [01:22] i don't have mailq [01:22] it tells me what package its in [01:23] apt-get install postfix [01:24] in the gui how can i select ok? [01:24] don't use a guo [01:24] don't use a gui [01:24] i mean i use terminal [01:25] but it looks like a gui [01:25] Package configuration [01:27] nvm [01:27] it was TAB + Enter [01:33] thanks everyone [01:33] especially RoyK [05:51] does anyone have cifs automounting working on 16.04? [05:53] i can successfully `smbclient -gL //server.fqdn/` with a password at commandline, but as soon as I use "-k" like auto.smb uses it throws errors === athairus is now known as afkthairus [08:36] coreycb, ok poked neutron-vpnaas, builds OK now [08:36] coreycb, nova - needs microversion-parse, heat - needs monascaclient [08:36] new deps, not in archive... [09:11] coreycb, switching to merge-mode=replace makes sense to me - updates made... [09:11] coreycb, poking at liberty failures now [09:12] coreycb, neutron/wily failure test failure looks genuine; glance liberty failures consistent across trusty and wily. [09:13] coreycb, I've also shoved the dh-python update into the SRU queue for Xenial - that will unblock most xenial things and the trusty/mitaka failures... [09:26] coreycb, reverting merge-mode for now - not supported on trusty [10:58] coreycb, ddellav: do you think major version matching might be a good idea for charm-helpers? [10:58] I really hate having to update for x.1's [10:59] hmm although that won't work for 20XX.X versions... [10:59] grak [11:20] coreycb, wedged the dh-python SRU into the openstack-ubuntu-testing PPA's to unblock branch builders... === JanC is now known as Guest47278 === JanC_ is now known as JanC [11:36] coreycb, will need to sru a glance-store point release for liberty [11:45] jamespage, ddellav, yes major version matching would be nice for charm-helpers [11:51] jamespage, thanks for all the updates, I updated the spreadsheet to track some of these and making a card for glance-store. [12:01] Morning! Does anyone have a page showing the steps to adding Ceph OSD's to an already deployed cluster with JUJU? Is it as easy as adding the drives to the server and rescanning using ceph-osd charm? [12:05] smoser, hey i have questions about config-drive metadata, networking, and static networking [12:06] i have provided a valid /etc/network/interfaces.d/enc1000.cfg with static network configuration... [12:06] and cloud-init ended up writing dhcp auto for the enc1000 interface in the /etc/network/interfaces.d/50-cloud-init.cfg [12:07] providing my own 50-cloud-init.cfg in the config-drive did not do a thing - cloud-init would still overwrite it with dhcp auto [12:07] i guess the only thing that worked was to provide /etc/networking/interfaces full stop. but that is sad =( [12:08] for networking json.... i did not find enough keys in it to configure static network configuration as needed either. [12:08] so questions [12:09] smoser, is it possible to use interfaces.d in xenial and trump 50-cloud-init.cfg? should i be recommending to the cloud provider to ship a straight up /etc/networking/interfaces ? [12:32] coreycb, pointer to spreadsheet? I can update myself... [12:33] jamespage, sent it to you [12:33] jamespage, thanks [12:44] coreycb, ta [12:54] coreycb, oh - sorry - I see you where already looking at nova [12:54] coreycb, I pushed my changes to the git repository... [12:55] my bad [12:55] now I can see the list that's great! [12:55] coreycb, the babel update may be ignorable for now [12:55] coreycb, heat wanted it as well, but built and worked ok without it... === vagarwal_ is now known as vagarwal [12:58] jamespage, no problem :) ack on babel === selinuxium_ is now known as selinuxium [13:01] Hi, any idea for partitioning a single machine for openstack testing ? There's a doc for automated cloud install but it does not recommand anything === jgrimm-afk is now known as jgrimm === habibi is now known as sexytime [13:40] xnox, there are some issues with config drive providing networking configuration right now. [13:40] if you do not want cloud-init to set fallback networking (write 50-cloud-init.cfg) then you have to disable cloud-init networking entirely. [13:40] i do not expectt that to change, but want to fix the config drive networking scenario in short order. [13:59] dunaeth, openstack deployed in LXD containers on a single machine? [14:04] coreycb, where I need to rev a dependency to fix a daily builds issue, I've been backporting a version to the ppa under ~openstack-ubuntu-testing as well [14:04] coreycb, glance-store being an example of that [14:07] jamespage, I guess that would mostly only be the case for stable releases, since it'll take longer to get deps uploaded to the archive for them [14:07] coreycb, yes [14:07] I think that's OK [14:07] coreycb, for dev, I've been uploading and then backporting straight away using backport_package - that places into -staging and the trunk testing ppa's [14:07] jamespage, it makes sense, for example I think glance-store dep probably should wait for the point release of glance before SRUing it [14:08] coreycb, Iagreed [14:08] jamespage, ok makes sense [14:09] coreycb, I think the mitaka build failures should be OK now [14:09] just waiting for the queue to clear :-) [14:10] jamespage, awesome! === manu is now known as Guest99998 [14:26] hello ubuntu-server fails during installation with an error "modprobe -v usb-storage failed" [14:26] are there any workarounds? [14:27] i'm trying to install from a usb stick and i don't have a cd/dvd rom drive [14:30] nabukadnezar43: check the 'dmesg' output for clues as to why/if the module fails to load [14:31] TJ-: ok let me try [14:38] I am looking to test latest Mitaka release available in latest openstack neutron tag ( https://github.com/openstack/neutron/releases/tag/8.1.0 ) on Ubuntu . But the updates are not yet available in Ubuntu Cloud archive as packages. Does anyone knows, how often OpenStack updates added to Cloud Archive? [14:43] hey basilAB - I know :-) [14:44] basilAB, apologies for not responding to your email - I was just thinking about doing that [14:44] ah! you are here. [14:44] great :-) thank you! [14:44] basilAB, typically we sweep up any avaliable stable release in the first two weeks of the month, with the aim of getting them out into -updates by the end of the month [14:45] basilAB, you'll be interested in the tracking bug - https://bugs.launchpad.net/cloud-archive/+bug/1580674 [14:45] Launchpad bug 1580674 in neutron-vpnaas (Ubuntu Xenial) "[SRU] mitaka point releases" [Undecided,New] [14:45] basilAB, work should progress this week... [14:46] subscribed now and thanks for the schedule details. I will keep an eye. [14:46] although we are a little blocked by a related dh-python issue - trying to get that clear first... [14:47] apperantly "usb_storage" module needs to be signed [14:47] jamespage: do you have LP for newton too? [14:48] jamespage: since you are here, heard or any plans on adding 'octavia' lbaas addition to cloud-archive? [14:48] EmilienM, we don't generally bug track development releases... [14:48] basilAB, not in the short term no [14:48] okay [14:48] how do i sign the usb_storage module for secure boot? [14:49] EmilienM, if you want to sniff current master branches for newton for Xenial and Yakkety - https://launchpad.net/~openstack-ubuntu-testing/+archive/ubuntu/newton [14:49] that is the state of currently built master branches - its not complete - working some new dependencies... [14:50] jamespage: nice! please ping me when you feel like I can start testing it (asap) [14:51] EmilienM, I'd not want you to put that into a voting gate btw... [14:51] jamespage: ok [14:57] EmilienM, the gate for that PPA is 'it builds and passes its unit tests' ... [14:57] ok [14:58] jamespage: as soon as all packages are there, please ping me, I'll start testing it and report you feedback. [14:58] EmilienM, that would be nice - thankyou! [14:58] cool [14:59] coreycb, hmm - niggle with adding a newer dh-python to the sbuild environment [14:59] working on that now... [14:59] it gets installed by s-p-c so we can add the PPA, which contains the newer version... [14:59] grrr [15:00] jamespage, s-p-c? [15:01] TJ-: dmesg output didn't show anything relevant but i tried probing usb-storage module manually. Got a "could not insert usb_storage required key not available" [15:01] error [15:02] nabukadnezar43: that sounds rather like a secure-boot issue; well checking the module signing key anyhow [15:02] software-properties-common [15:03] nabukadnezar43: the modules shipped with the distro should all be signed [15:04] TJ-: yeah that's weird [15:04] nabukadnezar43: which ubuntu release are you working with? 16.04 ? [15:04] 16.04 server amd64 [15:04] nabukadnezar43: using the -generic kernel, or -lowlatency? [15:05] jamespage, are you familiar with the get_component_config() error that several of the newton packages are hitting? [15:05] nabukadnezar43: it defaults to -generic but its always worth checking, I've noticed some differences with the -lowlatency as regards signing, though I forget what I did notice right now :) [15:05] TJ-: i have no idea, i haven't changed anything [15:05] jamespage, I'm looking at keystone and cinder for newton btw [15:12] nabukadnezar43: I can't see any obvious bug reports with a similar problem. But, as its the installer I'm wodering if the installer has kernel version A, and during the chroot installation of the latest kernel version B it tries to modprobe the version B usb_storage module, which would upset version A kernel I think (because the kernel uses different signing keys per build if I recall correctly) [16:16] hey guys [16:17]