[02:20] still having a problem... with a second NIC on my server [02:21] I added emp0s8:`n addresses: [192.168.1.1/24] to my 01-netcfg.yaml file [02:21] and ran sudo netplan apply [02:22] and sudo ip link set enp0s8 up [02:22] but, ip addr still doesn't show the ip4 address for emp0s8 === SmokinGrunts is now known as SG_Sleeps [07:06] Good morning [08:08] late hi everybody [11:45] cpaelzer: hi, iproute2 sru has been accepted, now we need to click on the reds in the excuses page [11:45] :/ [11:46] let me check that they are the same errors we saw before in bionic [11:48] who is chrony's maintainer? [11:48] he/she is bound to have seen these frequent dep8 test failures [12:06] Hi! Anyone has a guide to how I could achieve Dropbox style server without using owncloud or nextcloud? [12:17] ahasenack: ping me once you checked which ones seem flaky tests ok? [12:59] cpaelzer: systemd dep8 errors in s390: http://autopkgtest.ubuntu.com/packages/s/systemd/artful/s390x [12:59] mine is at the top (iproute), but others have failed in the same way [13:00] FileNotFoundError: [Errno 2] No such file or directory: '/boot/grub/grub.cfg' is the error [13:00] any idea about that? [13:01] it seems the systemd-fsckd test was skipped in the one lonely success [13:01] "systemd-fsckd SKIP Test requires machine-level isolation but testbed does not provide that" [13:01] ahasenack: ther eis no grub on s390x [13:01] and never will be [13:01] like 640k will be enough forever [13:02] there seem to be two "threads" writing to stdout in the failed test [13:02] I see lines like [13:02] (Reading database ... 95% [13:02] with "Setting up util-linux (2.30.1-0ubuntu4.1) ..." in between [13:03] let's see why that test isn't being skipped [13:03] ahasenack: I re-triggered the others, but the two s390x issues need to be resolved or skipped [13:04] well firejail might ahve been a race with another upload with some luck [13:04] or other out of date-ness [13:04] firejail failed like that before [13:04] but let me get to that in due time [13:05] ahasenack: once the others re-ran you can check then [13:05] ahasenack: if no others are left ask for overrides in #ubuntu-release [13:07] FYI ahasenackthese will eventually go into http://bazaar.launchpad.net/~ubuntu-sru/britney/hints-ubuntu-artful/changes [13:09] cpaelzer: the s390 tests were always in a vm, right [13:10] xnox: around? [13:11] ahasenack: no [13:11] ahasenack: they were in a container up until recently [13:11] ahasenack: which might be why they now are considered regressiosn [13:11] ah, that explains it [13:11] I didn't know you could do containers in s390 [13:11] ahasenack: that is what I fixed a few of last week [13:11] just as well as everywhere [13:12] so now that it's a vm, the machine-isolation constraint works and the test is run [13:12] but it uses grub, and that fails [13:12] so I need to skip that test in s390 [13:12] sounds reasonable? [13:13] * ahasenack looks for the dep8 spec [13:18] "Such specific HW need seems rare and there is no e.g. autopkgtest feature to limit Architectures." [13:19] probably need to add the skip to the test itself then, have it return a fake success [13:28] cpaelzer: I'm looking at your dpdk dep8 https://bugs.launchpad.net/bugs/1551158 fix [13:28] Launchpad bug 1551158 in dpdk (Ubuntu) "DPDK dep8 tests failing on non supported platforms" [High,Fix released] [13:28] you added debian/tests/check-dpdk-supported-arch.sh and you source that in the tests [13:28] but you also added arch-specific bits to the depends line [13:28] Depends: dpdk [amd64 i386] <-- [13:28] ahasenack: any ETA on the my MySQL merge review please? [13:29] ahasenack: yes and yes [13:29] rbasak: gonna start after I solve these iproute2 migration issues [13:29] ahasenack: the arch qualifier will ensure it doesn't run at all [13:29] as it doesn't qualify [13:29] cpaelzer: is the latter necessary? Without the former, the arch qualifier would just lead to a failed test? [13:29] ahasenack: no problem thanks! [13:29] the checker is mostly if even on an arch there are needs like cpu features [13:30] ahasenack: it is a double cahnce for error [13:30] ahasenack: without arch qualifier it will try to install and might fail [13:30] ahasenack: so you have to mark where you expect THE INSTALL to work [13:30] ahasenack: of the package [13:31] cpaelzer: the dpdk package does not exist in these other arches? [13:31] ahasenack: only afterwards the script will run and you can sort out and skip tests [13:31] yes [13:31] ok, so you also had an install failure that you are fixing here [13:31] yes [13:32] ok,thx [13:32] and check-dpdk-supported-arch.sh then does any in depth checks [13:32] like cpu features [13:32] or experimental arches like for a while ppc64el had the packages but was not meant to work [13:32] well that sounds bad [13:32] it worked [13:33] but was meant to be experimental/tech-preview [13:34] ahasenack: the story even went further, until s390x had KVM execution the isolation-machine blocked it [13:34] right [13:34] that's my case [13:34] it started failing about 3w ago, with several packages [13:35] I mean, other packages that triggered the systemd dep8 test suite [13:35] yep [13:35] I'm filing a bug and putting up an mp for it [13:35] I wonder how the bionic upload passed (of iproute2) [13:36] ahasenack: for the case to complete the story the final change then was https://gerrit.fd.io/r/gitweb?p=deb_dpdk.git;a=commitdiff;h=b179808726394c63b97747b31ca603392c182168 [13:36] because since KVM exec it ran into said package-install-issue [13:36] as we didn't have the arch qualifier on that yedt [13:36] can't you negate an arch there? [13:37] I don't know [13:37] sry [13:37] ok [13:44] cpaelzer: hm, "zipl" is an s390 thing, no? [13:45] * ahasenack looks around for didrocks [13:46] ahasenack: yes [13:46] ahasenack: zipl is the lilo of s390x [13:46] I think this test was meant to work on s390 [13:46] I might be out of my depth here then [13:46] if platform.processor() == 's390x': [13:46] enable_plymouth_zipl(enable) [13:46] else: [13:46] enable_plymouth_grub(enable) [13:46] yeah, there is some intention here [13:47] * ahasenack hops on an s390 to check what platform.processor() returns [13:47] it's correct [13:48] it is [13:48] ok, I need to actually run this test there then [13:52] cpaelzer: am I supposed to be able to run autopkgtest-buildvm-ubuntu-cloud on s1lp5? Or do I need to use nested vm? [13:52] ubuntu@s1lp5:~/andreas$ autopkgtest-buildvm-ubuntu-cloud -r artful -o adt-images [13:52] ERROR: no permission to write /dev/kvm [13:59] ahasenack: I thik lp4 is the one we share [14:00] ahasenack: but long story short no [14:00] ahasenack: the tests won't work [14:00] there is a lot of console magic in autopkgtest which doesn't apply [14:01] ahasenack: create a VM with uvtool, then run the test in that VM (without the autopkgtest around it) [14:01] only go the last steps to try inside if you really really need it [14:04] it must be platform.processor() returning something else over there [14:04] yep [14:04] maybe it fails in a VM? [14:04] the test clearly ran plymouth_enabled = 'splash' in open('/boot/grub/grub.cfg').read(), which is only in enable_plymouth_grub() [14:04] yeah, let's start ismple. Bring up the vm and run that platform.processor() [14:05] doing that atm [14:05] cpaelzer: hm, there is no uvt-kvm in that s1lp5 host, should I switch to that lp4 one you mentioned? You gave me access to lp5 once upon a time, maybe before lp4 was ready for us? [14:06] lp5 is mostly mine for the more sinister experiments [14:06] there used to be uvt-kvm, since I ran it before there [14:06] ah, ok [14:06] lp4 is meant to be the somewhat stable shared host [14:06] better remove me from lp5 then :) [14:07] it is s390x on a KVM guest as well [14:08] trying to run the full test [14:08] "s1lp4 purpose: jenkins node" :) [14:08] s1lp3 seems to be the one to use [14:09] I wrote it in the wiki [14:09] yep s1lp3 [14:20] ah, found it [14:22] it's a fix that went into bionic [14:22] New changelog entries: [14:22] * systemd-fsckd: Fix ADT tests to work on s390x too. [14:22] somehow I missed that changelog entry [14:23] cpaelzer: it's a bug in artful's package, fixed in bionic. We probably don't sru dep8 fixes, or do we? [14:29] ahasenack: we soemtimes do sometimes not [14:29] depends on the case [14:30] but systemd uploads are grouped by xnox anyway [14:30] you know he collects a bunch and groups them for tests [14:30] so he might have a plan or nack already [14:30] I guess you are safe to ask for an override on the current version thou [14:30] ahasenack: ^^ [14:31] thanks, I'm asking in #ubuntu-release [14:31] if you want you can explain so in a bug, release team members like to reference something with more context [14:31] as it is just a lin in the britney hints [14:33] cpaelzer: I have a bug, can you accept the artful nomination? https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1736955 [14:33] Launchpad bug 1736955 in systemd (Ubuntu) "dep8 test systemd-fsckd fails on s390" [Undecided,Fix released] [14:36] ahasenack: I don't think we'd usually SRU a test fix on its own, but bundling one with an SRU is absolutely fine. [14:37] sounds reasonable [14:37] approved [14:37] it is correct to have that bug task [14:37] and you can refer to it for the override [14:38] cpaelzer: thx === nitemare is now known as trobotham [15:51] cpaelzer: one more task, zesty is also affected: https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1736955 [15:51] Launchpad bug 1736955 in systemd (Ubuntu Artful) "dep8 test systemd-fsckd fails on s390" [Medium,Triaged] [15:51] xenial and trusty are fine (no systemd dep8 tests being run in their migrations) [16:04] for the no_proxy environment variable, i often see addresses and hostnames. the docs say categorically that hostnames should be used. also, i often see both address and name for the local system: 127.0.0.1, localhost. is this necessary? [16:07] pmatulis: surprisingly I've seen many cases where an app would send a request to localhost via the proxy [16:07] I don't know why that's not the default [16:19] no_proxy has always seemed partially implemented to me [17:05] I have the following lines in my preseed postinstall script to upgrade and configure ansible to the latest version. [17:05] https://gist.github.com/boxrick/ae85da1eedd485930a37a3dfd6e08329 [17:05] But I wish for this to happen in the preseed itself and not the post install [17:05] 3Any ideas? [17:06] ahasenack, thanks [18:34] Does this make any sense? https://gist.github.com/asbachb/9fceeb1d0a00114eec31c6af82ca9805 [18:40] Is 2001:470:4242:1042::1/64 the same as 2001:470:4242:1042::1/56 ? [18:40] m15k: yes, ifup/down will only work with interfaces defined in /etc/network/interfaces [18:41] ahasenack: Any idea howto remove the interface? [18:41] m15k: what ubuntu is this? [18:41] 16.04 [18:42] m15k: does "lxc network list" list that interface? [18:43] ahasenack: yes. but "lxc network delete lxcbr0" results in "error: not found" [18:43] is managed "no" for it? [18:43] yes it's no [18:45] that interface was created when you first installed lxc (not lxd: then it would have been lxdbr0). Are you sure you want to remove it? Do you use lxc or lxd? [18:45] Yes I am sure. I think I created it manully via brctl [18:46] does /etc/default/lxd or /etc/default/lxc (or a name like that) reference it still? [18:47] ahasenack: I think that's it. lxc-net bridge was enabled and referenced to lxcbr0 [18:48] you can probably change its details in that /etc/default/ file [18:48] ahasenack: thanks. that was the right hint! :) [18:49] cool === LudusLight is now known as Ludus [18:53] ahasenack: Are you familiar with lxd? === Ludus is now known as LudusLight [18:53] m15k: somewhat [18:54] I use it a lot [18:54] I wonder what "Would you like LXD to NAT IPv6 traffic on your bridge?" actually means [18:55] Hello! I am currently using the following line in preseed on Xenial 16.04 LTS d-i base-installer/kernel/altmeta string hwe-16.04 [18:55] So I have have the more up to date kernel [18:55] I know what NATting is, but I'm a little bit unsure what's the difference in ipv6 context. [18:55] However this causes all sorts of inconsistencies within my preseed postinstall chroot environment where I need correct libraries [18:56] Is there any way to install *just* the new kernel rather than do the thing it seems to where it installs the old one then updates it later? [19:00] m15k: well, it depends if you have global addresses in your lxds or not [19:01] if you don't, and you want to use ipv6 to reach the internet from that container, then you will probably need ipv6 nat, but also a global ipv6 on your host [19:01] ahasenack: So when I've a public ipv6 subnet I should disable NAT? [19:01] if your containers get a slice of that and have global addresses, probably yes [19:01] I have never natted ipv6, tbh [19:01] I just get one /64 [19:02] ahasenack: You assign a public ipv6 to your containers? [19:02] no [19:02] I don't use ipv6 in them [19:03] I currently try that. Because of that I play a little bit around with these bridges. [21:39] When I type "resolvconf -u" there are dns servers in /etc/resolv.conf that are not configured in "/etc/resolvconf/resolv.conf.d" any ideas how they get into the generation? [21:39] Probably by dhcp [21:58] powersj: congrats! [21:58] rbasak: thank you :) [22:15] oh, it happened? [22:15] nice [22:54] who know my server can access server sysadmin? [22:55] does sysadmin of server can access my server? [22:55] I mean files on my server [22:57] root can generally access anything that is not encrypted. [22:58] and if the data is ever decrypted on the server, root can access that too.