[test review] snapper_undochange fails to create first snapshot within time of 30s
test fails for no apparent reason so indication is that it might be openQA performance limitations
occured at least twice:
H1. test timeout of 30 seconds is too strict to ensure a snapshot can be created under all circumstances
H2. a snapshot sometimes can not be created at all, waiting for longer would not help
H3. ppc64le specific
#1 Updated by sebchlad over 6 years ago
about snapper_undochange failure: i remember that this test was failing also time ago, is a performance issue.
i would just correct the test, and remove the wait_serial
instead of script_run("test -f $snapfile || echo \"snap-ba-ok\" > /dev/$serialdev", 0);
just test with assert_script_run(test-f $snapfile)
#2 Updated by okurz over 6 years ago
- Subject changed from [test review] snapper_undochange test failes for no apparent reason to [test review] snapper_undochange fails to create first snapshot within time of 30s
- Description updated (diff)
- Category set to Enhancement to existing tests
please reference links to failing test runs for tickets like these: I assume you found the issue in https://openqa.suse.de/tests/441169, this is where you referenced the current ticket.
It also failed in a previous run, Build 1579: https://openqa.suse.de/tests/409989#step/snapper_undochange/5
When it succeeds it normally takes way shorter than the timeout of 30s, e.g. in https://openqa.suse.de/tests/412046/file/autoinst-log.txt only about 3s, also here https://openqa.suse.de/tests/434681/file/autoinst-log.txt and here https://openqa.suse.de/tests/409021/file/autoinst-log.txt which does not confirm H1 (w/o rejecting it).
#12288#note-1 does actually not concern this specific step in the test but a different one so it should have no effect on this specific issue.
#3 Updated by okurz over 6 years ago
- Status changed from New to Resolved
- Assignee set to okurz
Author: Oliver Kurz email@example.com
Date: Tue Jul 26 09:10:48 2016 +0200
Fix snapper test if snapshot is below first page We need to go down the first page, e.g. with 'pgdn' to ensure the last snapshot is visible before looking for it. Should fix errors observed as in https://openqa.suse.de/tests/485232#step/yast2_snapper/31 . Simplify the structure, replace sleep and basically bring the snapper test into a current state as how we write tests to prevent wasting needless wait time. Verification run: http://lord.arch/tests/2042