Using this patch:
diff --git a/test/gvfs-test b/test/gvfs-test
index 4537d8a5..c82e13a8 100755
--- a/test/gvfs-test
+++ b/test/gvfs-test
@@ -278,7 +278,7 @@ class GvfsTestCase(unittest.TestCase):
ml = GLib.MainLoop()
-mount.unmount_with_operation(Gio.MountUnmountFlags.NONE,
Unfortunately the timeout bump didn't fix the problem:
==
FAIL: test_anonymous_api_user (__main__.Ftp)
ftp:// anonymous (API with user)
--
Traceback (most recent
We need a way to reproduce this locally, so that we can get a shell and
investigate what is going on. So far I tried locally with amd64 and i386
vms, and the tests always pass. I guess I really need a ppc64el one.
--
You received this bug notification because you are a member of Ubuntu
Bugs,
I included that patch in 1.36.0-1ubuntu1 just uploaded, but I forgot to
close this bug with that upload, sorry.
Let's watch the tests and see how they go.
** Changed in: gvfs (Ubuntu)
Status: New => Fix Committed
--
You received this bug notification because you are a member of Ubuntu
After increasing the timeout, I ran the tests 5 times each on i386 and
amd64.
https://paste.ubuntu.com/p/tXS2dnzYVy/
It passed every time - so it looks as if this might at least help. I
just filed https://bugzilla.gnome.org/show_bug.cgi?id=794487 to see if
we can get this merged upstream.
**
I don't know if the timeout is borderline short, but the unmount definitely
failed. Logs further down (sorry, these logs are interleaved and have no
timestamp) show that unmount failed because the filesystem was busy:
ftp: Queued new job 0x1002d932210 (GVfsJobCloseRead)
ftp: <- 0 -- 226
This is the first test that fails:
==
FAIL: test_anonymous_api_user (__main__.Ftp)
ftp:// anonymous (API with user)
--
Traceback (most recent call last):
File
Oh, look at that unmount_api() docstring:
def unmount_api(self, gfile):
'''Umount a mounted Gio.File using the Gio API
This times out after 5 seconds.
'''
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
s390x and arm skip the test so all green there isn't worth anything.
The others fail in ~30% of the cases I'd say:
http://autopkgtest.ubuntu.com/packages/gvfs/artful/ppc64el
http://autopkgtest.ubuntu.com/packages/gvfs/artful/i386
http://autopkgtest.ubuntu.com/packages/gvfs/artful/amd64
--
You
The errors read like networking or concurrency breaking it.
I agree to have the bug to debug and improve the case.
For now I re-triggered the one hanging test on artful.
The release Team can mark dep8 tests as flaky to skip them blocking a
migration, but that would prereq an analysis of the case
10 matches
Mail list logo