[03:26] hi all. does anyone know how to get a load average with `sar' command? [03:36] Hello, need some assistance with mdadm / fstab . i have 18.04 server, sda for system, sdb and scd as raid0 md0 device. the first problem is after reboot md0 became md127, is it ok? or i need start over? [04:14] michael2: no, but here's some reading that you may enjoy http://www.brendangregg.com/Articles/Netflix_Linux_Perf_Analysis_60s.pdf http://www.brendangregg.com/USEmethod/use-linux.html [06:16] I have setup a server using the 'ubuntu-server' meta package... I am trying to get netplan to create a bridge with a static IP... the method used in every doc. I have found doesn't seem to work... can anyone help here. The "addresses: [10.20.0.1/16]" doesn't seem to do anything === lifeless_ is now known as lifeless === zul_ is now known as zul === miguel is now known as Guest62304 [10:20] cpaelzer: did you get an error running "git ubuntu merge start" on your ntp merge? [12:10] nacc: looks like dpkg-parsechangelog eliminates duplicates, but debian.changelog.Changelog does not. The assertion check caught this for ntp. I have a minimal test case. Additionally I found that the assertion check has an edge case bug when trying to prepare a test case. [12:11] So I now have two bugs: one where the assertion check fails to assert, and the underlying issue when it does assert due to a duplicate. [12:11] This is a bit mind bending so I thought I'd share. I'm wondering how to proceed. [12:11] Use case: git ubuntu merge start fails on ntp because of a historical changelog error [12:11] In that case it doesn't really matter that the two implementations mismatch [12:12] However this kind of difference is why we wrote the assertion check in the first place [12:12] Does this difference affect our algorithm? [12:40] Question: is fail2ban supposed to work out of the box? (without any configuration) [13:16] neildugan: have you looked at https://netplan.io/examples#bridging ? [13:17] neildugan: otherwise you should paste your entire netplan config so we can look at it [13:45] o/ morning [13:46] I am REALLY starting to like netplan :) [13:52] It's awesome, isn't it! [13:53] yes [13:54] just took some getting used to [13:58] hi #ubuntu-server. i have another fun problem. i just tried resizing my ubuntu's / partition (ext4) using centos 7, but resize2fs failed because the filesystem apparently has "unsupported feature(s)" [13:58] any ideas of which feature(s) could/would be causing this problem? [13:58] (not sure if i should be asking here or a centos channel tbh...but i assume this place will know more about what features ubuntu uses for ext4 by default) [14:00] LambdaComplex: that's a known thing, let me find the bug [14:01] sdeziel: i found something on the centos mailing list but it devolved into bikeshedding over updating for reboots [14:01] and i don't think a conclusion was ever reached wrt the original topic [14:02] LambdaComplex: https://lists.ubuntu.com/archives/ubuntu-server/2018-March/007671.html [14:03] ext4 feature 'metadata_csum' [14:03] gpartd [14:04] LambdaComplex: that said, I always resize the root FS live [14:04] sdeziel: lol, you said that right as my coworker messaged me something about metadata_csum [14:04] so i'd say that's definitely the thing [14:05] LambdaComplex: so if the partition boundary is moved already (CentOS was probably able to do that part), I'd boot on 18.04 then run "resize2fs /dev/fooX" [14:05] live resize never failed me [14:22] I'm upgrading from 16.04 to 18.04 and wrestling with systemd-resolved. I need to forward requests under one FQDN to a local DNS server, all other requests should go to the DNS acquired via DHCP. I had this working with dnsmasq previously, but doesn't work after upgrading. Is DNS forwarding per FQDN/domain possible with systemd-resolved? Or do I need to disable it and use dnsmasq only? [14:23] If I do disable it, how can I ensure that dnsmasq is setup with the DHCP-acquired dns-server (which I assume may change from time to time)? [14:25] sandstrom: systemd-resolved supports per domain forwarders [14:25] dunno how to configure it that way though [14:25] @sdeziel sounds awesome :) [14:26] @sdeziel I've looked through the docs, haven't found anything [14:28] @sdeziel there is support for search domains, but that is something else [14:31] sandstrom: hmm, maybe I was wrong. systemd-resolved supports per-link resolvers/forwarders [14:31] sandstrom: not sure anymore about the per domain [14:32] @sdeziel Yeah, I didn't find anything in `man resolved.conf` at least [14:32] @sdeziel guess I'll just have to bypass systemd-resolved entirely [14:33] sandstrom: have you tried a systemd specific channel? [14:34] @sdeziel know of any? [14:34] I'll try in #systemd [14:35] sandstrom: that's where I'd go ;) [14:35] cyphermox: my coworker wants to know how he can make netplan "put up a bridge containing only a vlan interface with no addresses assigned?" [15:35] WAT ? [15:47] LambdaComplex: right not, you can't, unless you add another .network file setting ConfigureWithoutCarrier=yes [15:48] I'm fixing that very very soon, working on a new release right now [15:51] cyphermox: that's good to hear cause I was also missing that feature [15:59] rbasak: I hit all sort of errors during merges [15:59] rbasak: none that I'd not have reported already in one or another form [15:59] rbasak: not sure if the one I hit on ntp was on merge start or later [16:00] rbasak: I quite often have to do the rebase again, which seems due to old merge tags being around [16:00] I haven't proven yet what the source is e.g. how to trigger [16:00] but in those cases I do the rebase myself, and have to take out ALL history until it starts at the right place [16:00] so something at the parenting seems to be mixed up [16:01] rbasak: mostly I hit the "error has to be clean" when creatin logical tags [16:01] in those cases I create it myself [16:03] rbasak: do you have a log? [16:10] nacc: I can produce one but I'm confident I know exactly what's going on [16:10] nacc: I intend to write it all up in a bug [16:10] With minimal reproducers etc [16:11] nacc: minimal test case: http://paste.ubuntu.com/p/6D3VpsshrP/ [16:11] nacc: minimized reproducer: https://pastebin.ubuntu.com/p/XZsc8g2Frr/ [16:12] nacc: in the assertion, "zip" throws away the end of the longer input, so the minimal test case needs the fourth changelog entry to trip it up. I think that's a bug in the assertion code. [16:12] I've come to the conclusion that it doesn't matter for the importer, but I'll double check. [16:13] Since the changelog parent is defined as the first occurance that matches, the issue of a second match never comes up [16:13] So I think what I'll do is fix and relax the assertion [16:13] If the assertion dedupes both input lists before asserting, then the assertion will still pass and detect any other mismatch problems [16:14] While explicitly accepting this particular class of mismatch [16:16] To be clear, that means I think there are two things that need fixing. [16:16] 1) Relaxation of the assertion to allow a duplicates-exist vs. deduped difference assuming that we all conclude that this is safe. [16:17] 2) Fixing of the assertion in the zip truncation case that causes the comparison to fail to assert with only the first three changelog entries in my example. [16:17] rbasak: ack [18:58] hi, i got a question. Ubuntu server 18.04. I see /etc/network/interfaces is gone. I'm reading about netplan. I got a file on /etc/netplan called 50-cloud-init-yaml. Is the only file in that directory. It has a comment that says that changes on that file will not be persistent. I want to fix a static ip address. [18:58] Do i have to delete that file and create a new one with sudo netplan generate? [18:59] it's probably worth spending a few minutes skimming the website if this is your itnroduction to netplan https://netplan.io/ [19:00] will do sarnold [19:00] thnx [19:00] how do I check my ubuntu server version? [19:01] cat /etc/issue [19:16] sarnold: Added !netplan just now, for future reference [19:16] genii: thanks [19:16] np [19:25] rbasak: did have a thought about your test case [19:26] rbasak: 1) I'm not sure that Ubuntu will let you do that, if those are all actually published, as 3 would be published with different contents [19:27] rbasak: 2) if it is allowed we should see a reimport tag, and two distinct entries, but I think with my branch that should all work (when we see the second 3, we'd create a reimport tag) [19:28] granjero: you can modify that file, the mention that it's not persistent is essentially a lie. [19:28] or you can delete and create a new one, but you shouldn't use 'netplan generate'; use 'sudo netplan apply' instead [19:28] rbasak: meanwhile, did you figure out with kyrofa what is going on with 430? [19:29] genii: thanks! [19:32] rbasak: also afaict, ntp's `git ubuntu merge start` failure has nothing to do with this code? rather confused [19:32] I see a failed cherry-pick [19:34] oh i know why this happened, I think [19:34] after the reimport, our merge logic might need adjusting [19:35] FYI I am VERY happy with netplan [19:53] I'm not [19:53] I'd be if I could fix these autopkgtests :) [19:54] Ussat: if your config is more than just dhcp or static addresses though I'd be very interested to have a look; see if we can make that a generic example [20:05] cyphermox, thanks. it's confusing thats says that changes to the file will not be persistent. =P will bakup that file and create a new one "config.yaml" [20:05] excuse my english [20:06] it is a bit rusty [20:06] granjero: well, it's persistent unless the metadata changes, which depends on what kind of server it is [20:07] also confused about the indentation of the file. [20:07] is there a "tab" then 2 tabs an so on? [20:08] don't use tabs [20:10] spaces only [20:19] cyphermox, the name of my ethernet interface comes from the comand ip? [20:19] ip a? [20:19] yes [20:19] ip link [20:20] great. thanks [20:21] nacc: o/ [20:21] nacc: git ubuntu merge start --tags-only currently fails on ntp [20:22] nacc: the root cause I believe is that dpkg-parsechangelog output mismatches the parsing of debian.changelog.whatever [20:22] nacc: and that fails the assertion [20:22] The reason for the mismatch is that in the ntp changelog currently in Ubuntu, a changelog entry is duplicated [20:22] Like it is in my test case [20:24] nacc: I haven't checked but I suspect that there weren't actually two uploads or anything. It's a changelog transcription error from the uploader probably. [20:25] Or else it's intentional and represents how the maintainer decided to make something in experimental appear linear, etc. [20:35] cyphermox, worked like a charm! [20:35] thank you very much [20:35] sarnold, thank u too [20:48] [ 10.020680] mlx4_en: enp2s0: Link Up [20:48] [ 499.936066] EXT4-fs (sda2): mounted filesystem with ordered data mode. Opts: (null) [20:48] well thats not good :p [20:48] that is / by the way [20:50] rbasak: uh ... it works fine here [20:50] rbasak: what version of the snap? [20:50] rbasak: r430 is unusable, as you know, so i'm manually on r429 [20:50] rbasak: what changelog entry do you see being duplicated? [20:51] rbasak: i'm fairly sure the code you're talking about is completely unused by merge [20:57] rbasak: fresh clone, merge start --tag-only worked fine with r429 [21:00] now the rest of `merge start` does not ... but that is what i was referring to earlier [21:00] hm. it is blocking in initramfs. could be entropy maybe [21:03] hello nacc [21:03] Blueking: hey [21:03] sup ? [21:04] Blueking: ... nothing? you pinged me :) [21:04] :) [21:04] got mine xeon E3 'server' fixed :) [21:04] Blueking: nice! [21:05] had to perform bios flash with ME disabled [21:30] strange, it is modprobe ext4 that is blocking?? [21:31] and it unblocks the second I send it a alt+sysrq (with no sysrq command) [21:31] hmm, isn't that always builtin on Ubuntu kernels? [21:31] sdeziel: yes. it IS strange ;) [21:32] now if I could figure out how to give it a alt-sysrq-l through this POS java thing [21:33] of course if it unblocks when I get to alt-sysrq that wont work ha [21:34] hitting the keyboard until crng inits does nothing, but alt-sysrq immidiately unblocks it. [21:35] * trippeh_ confused [21:35] it is a Atom C3758, fwiw [21:36] and on cosmic (pretty sure I saw it on bionic too) [21:41] rbasak: heh, it will work properly with our non-merge upload tags [21:42] rbasak: but is pretty broken with our merge ones, and i need to think of how it ever worked [21:50] hm on a hunch I started blacklisting drivers during initramfs [21:50] blacklisting ixgbe seems to fix it [21:51] ..but now it locked up on shutdown hrm ;) [22:10] ok I've pretty reliably moved the hangs from boot to shutdown. progress? :P [22:12] heh, now you just need a reliable way to yank the power .. [22:15] the bios is confused about the ixgbe now. there might be a pattern.. [22:16] sarnold: https://imgs.xkcd.com/comics/hard_reboot.png [22:17] genii: hehehe :) [23:06] rbasak: cpaelzer: ok i see why merge start failed [23:06] pkg/upload/1%4.2.8p10+dfsg [23:06] bah, pkg/upload/1%4.2.8p10+dfsg-5ubuntu2 [23:06] does not have an appropriate import-tag as a parent [23:07] it was based off a point where we had the devel branches separately being merged throughout the history [23:08] thus, `merge start` doesn't see those commits (they are not in the ancestry list) and so the cherry-pick fails [23:08] rbasak: that's the only bug i see with ntp currently [23:09] rbasak: i *think* our new algorithm would reject this upload tag