testing/os-autoinst-distri-rocky-migrated
10
0
Fork 0
Code Issues 12 Pull Requests 1 Packages Projects Releases Wiki Activity

Handle update build not having any packages for target arch

Browse Source 
openQA sometimes winds up testing an update that doesn't have
any packages for x86_64 (or aarch64). The most common case is
s390utils, which is on the critpath but only has packages for
s390x. I would ideally like to skip scheduling entirely if the
update has no packages for the arch we're scheduling on, but
sadly that involves using the Koji API which is XML-RPC and I
don't really want to deal with that again. This deals with it
at the test level instead, by checking the error message if
`koji download-build` fails and carrying on if it's the "no
packages for this arch" error. That means if the update has no
packages at all for our arch we're not really testing anything,
but that's better than a bunch of false failures, I guess.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
This commit is contained in:
Adam Williamson 2020-10-07 09:29:18 -07:00
parent ea0dc63fc3
commit 1fdf7cd567
1 changed files with 7 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
lib/utils.pm
View File

@ -543,7 +543,13 @@ sub _repo_setup_updates {
if (get_var("ADVISORY_NVRS")) {
# regular update case
foreach my $nvr (split(/ /, get_var("ADVISORY_NVRS"))) {
assert_script_run "koji download-build --arch=" . get_var("ARCH") . " --arch=noarch $nvr", 600;
if (script_run "koji download-build --arch=" . get_var("ARCH") . " --arch=noarch $nvr 2> download.log", 600) {
# if the error was because the build has no packages
# for our arch, that's okay, skip it. otherwise, die
if (script_run "grep 'No .*available for $nvr' download.log") {
die "koji download-build failed!";
}
}
}
}
elsif (get_var("KOJITASK")) {