mwhudson | the build failed on arm64 because of course it did | 00:54 |
---|---|---|
mwhudson | retrying | 00:56 |
jawn-smith | mwhudson: Go 1.17 is resolving the segfaults | 02:13 |
mwhudson | jawn-smith: ok interesting. maybe we should file a ffe around changing the defauilt | 02:14 |
jawn-smith | I can do that! | 02:14 |
mwhudson | jawn-smith: do you think you have time to do some rebuild testing? | 02:14 |
mwhudson | jawn-smith: the way i've done this before is to create three ppas: | 02:15 |
mwhudson | 1) has golang-defaults with bumped default | 02:15 |
mwhudson | 2) is something like go-baseline-rebuilds | 02:15 |
mwhudson | 3) is something like go-117-rebuilds, and depends on the ppa in 1) | 02:15 |
mwhudson | then run copy-package $(reverse-depends -b src:golang-defaults) twice | 02:16 |
mwhudson | oof that's ~1750 packages now | 02:17 |
mwhudson | .oO(why does reverse-depends put its output on stderr not stdout??) | 02:17 |
jawn-smith | oof indeed | 02:18 |
jawn-smith | I shall make some ppas. Thanks for the tips. As much as I'd like them to run stuff overnight I'll likely start on that tomorrow, as it's gotten pretty late here | 02:18 |
mwhudson | fair enough | 02:20 |
amurray | hey folks, so I uploaded a security fix for squashfs-tools to impish-proposed and it is failing to migrate due to snapd autopkgtest failing with it - eg. https://autopkgtest.ubuntu.com/results/autopkgtest-impish/impish/amd64/s/snapd/20210831_232709_b9c56@/log.gz - is this because it got built against a different version of systemd than snapd did? any ideas how I can try and resolve this? | 02:36 |
amurray | I think the relevant part of the log is https://paste.ubuntu.com/p/QyWPDcYFqJ/ | 02:37 |
amurray | do I just hint this or should I try a no-change-rebuild or is there some other/better option? | 02:38 |
amurray | I am assuming I don't hint it since making things uninstallable doesn't seem the right thing to do :) | 02:39 |
jawn-smith | mwhudson I decided to do it anyway | 02:48 |
sarnold | hehehe | 02:51 |
amurray | hmm or is this perhaps due to different versions of glibc being in -proposed when snapd and systemd were built..? | 02:57 |
Unit193 | All I know is I keep getting emails which I can't do anything about. :( | 03:17 |
schopin | autopkgtest excuses question: I have a regression here https://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html#wireguard but I'm fairly certain it has been fixed in the netplan.io version currently in -proposed. Would it suffice to simply re-run the tests with the new version to unblock wireguard? (ignoring the glibc elephant in the room) | 10:45 |
sil2100 | schopin: yeah, we can try to re-run the tests with an additional netplan.io trigger | 10:47 |
sil2100 | schopin: I'll run that for you | 10:49 |
schopin | thx. Out of curiosity could you MP me the URL to check that I'd've gotten it right? | 10:50 |
ginggs | sil2100: schopin already passed https://autopkgtest.ubuntu.com/packages/n/netplan.io/impish/amd64 | 10:55 |
schopin | Great minds, and all that ;) | 10:58 |
ddstreet | mwhudson please, please, don't just cat patches together into a single quilt patch. one quilt patch per upstream patch, please. please. e.g. lp:1927161 | 12:51 |
=== genii-core is now known as genii | ||
jawn-smith | hmm I might be missing something, but why does golang-1.16 not seem to have autopkgtests? It seems like something that we would want tests for | 15:48 |
dbungert | May I get some assistance with retest clicks? amd64 passed with a similar retry. | 15:51 |
dbungert | https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=arm64&package=snapd&trigger=squashfs-tools/1%3A4.4-2ubuntu1&trigger=glibc/2.34-0ubuntu2 | 15:51 |
dbungert | https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=ppc64el&package=snapd&trigger=squashfs-tools/1%3A4.4-2ubuntu1&trigger=glibc/2.34-0ubuntu2 | 15:51 |
dbungert | https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=s390x&package=snapd&trigger=squashfs-tools/1%3A4.4-2ubuntu1&trigger=glibc/2.34-0ubuntu2 | 15:51 |
ginggs | dbungert: . | 15:53 |
dbungert | ginggs: ty | 15:56 |
dbungert | ginggs: one more, if you would | 16:10 |
dbungert | https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=ppc64el&package=snapd&trigger=squashfs-tools/1%3A4.4-2ubuntu1&trigger=glibc/2.34-0ubuntu2&trigger=golang-1.16/1.16.7-1ubuntu1 | 16:10 |
bdmurray | I got it | 16:11 |
dbungert | bdmurray: ty | 16:11 |
bdmurray | dbungert: back to this url you can't view it correct? https://autopkgtest.ubuntu.com/results/autopkgtest-impish-dbungert-vip-manager-armhf/?format=plain | 16:12 |
dbungert | it started working at some point | 16:12 |
dbungert | I even tried force-refreshes on the browser previously so I have no idea why that might have changed | 16:12 |
bdmurray | okay if you can't I've been informed that's a bug | 16:13 |
dbungert | it seems that golang-1.16.7 may not be enough to address the compile-time segfaults on ppc64el | 16:32 |
dbungert | bdmurray: more vip-manager fun? I have a prospective fix https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=armhf&package=vip-manager&ppa=dbungert/vip-manager-armhf&trigger=vip-manager/1.0.1-4ppa3 | 16:45 |
bdmurray | dbungert: clicked | 16:50 |
jawn-smith | dbungert: what makes you say that aobut golang-1.16.7? | 18:32 |
jawn-smith | ah the snapd fails | 18:34 |
jawn-smith | alright, I'll resume baseline retesting with 1.17 | 18:34 |
dbungert | is there a possibility we need more trigger packages? | 18:34 |
dbungert | we could try another all-proposed | 18:34 |
jawn-smith | Seems reasonable | 18:35 |
jawn-smith | we could also try some little packages with short tests | 18:35 |
jawn-smith | snapd is a bit of a beast | 18:35 |
jawn-smith | but some go packages like golang-github-ajg-form have pretty small test suites | 18:35 |
jawn-smith | let me whip up a URL | 18:35 |
dbungert | I'll sit tight then and let you ponder that ;) | 18:35 |
jawn-smith | Could any core dev give this a try? | 18:38 |
jawn-smith | https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=ppc64el&package=golang-github-sethvargo-go-fastly&trigger=golang-github-ajg-form/1.5%2Bgit20160822.523a5da-1.1ubuntu1&trigger=golang-1.16/1.16.7-1ubuntu1 | 18:38 |
jawn-smith | it's not all-propsed, I want to see how a little package does with go 1.16.7 | 18:38 |
jawn-smith | oof my keyboard battery is dying | 18:38 |
bdmurray | jawn-smith: I got it | 18:45 |
jawn-smith | thanks! | 18:45 |
jawn-smith | https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=ppc64el&package=golang-github-mattn-go-gtk&trigger=glib2.0/2.68.4-1&trigger=golang-1.16/1.16.7-1ubuntu1 | 19:36 |
bdmurray | are you just pasting random urls in here now? | 19:36 |
jawn-smith | Nope! I chose this one because it's showing very consistent segmentation faults with 1.16.5, which I then recreated locally | 19:37 |
jawn-smith | After recreating locally I upgraded to golang-1.16.7 and it started building | 19:37 |
bdmurray | okay, I'll click it | 19:37 |
jawn-smith | Thanks! | 19:38 |
jawn-smith | That one's also related to a proposed-migration task I volunteered for in the weekly meeting | 19:40 |
jawn-smith | I'm reasonably confident that Go 1.16.7 fixes the segmentation fault issue in ppc64el. However, the packages with the segfaults don't directly depend on golang-1.16. Instead, they depend on golang-defaults which points to 1.16. Because of that, the autopkgtests for those packages aren't automatically restarted. Is there a way to start them in bulk or does a list need to be put together of | 20:41 |
jawn-smith | tests to start manually? | 20:41 |
jawn-smith | On a related note, those that failed on ppc64el were almost all triggered by glibc 2.34 | 20:42 |
bdmurray | jawn-smith: are they still listed as regressions in excuses? | 20:43 |
jawn-smith | yes | 20:43 |
jawn-smith | bdmurray: they are listed as regressions blocking glibc | 20:44 |
bdmurray | there's this retry-autopkgtest-test script in ubuntu-archive-tools | 20:45 |
bdmurray | https://git.launchpad.net/ubuntu-archive-tools/tree/retry-autopkgtest-regressions | 20:45 |
jawn-smith | I see. The --log-regex flag seems particularly useful here | 20:47 |
bdmurray | well let me know if you come up with something you want me to run | 20:51 |
jawn-smith | bdmurray: https://paste.ubuntu.com/p/BtRDMsydmb/ | 21:10 |
bdmurray | jawn-smith: I'm pretty sure I submitted them all | 21:16 |
bdmurray | https://autopkgtest.ubuntu.com/running | 21:16 |
jawn-smith | Thanks! | 21:16 |
jawn-smith | yeah I see lots of them popping up | 21:17 |
jawn-smith | It looks like quite a few of those ppc64el golang tests have passed already \o/ | 21:23 |
=== genii is now known as genii-core |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!