Casey's fix worked, woot ! http://pulpito.ceph.com/yuriw-2019-11-08_14:54:42-smoke-nautilus-distro-basi... http://pulpito.ceph.com/yuriw-2019-11-08_14:54:32-smoke-mimic-distro-basic-s... http://pulpito.ceph.com/yuriw-2019-11-08_14:54:19-smoke-luminous-distro-basi... There are two issues known to me that need to be fixed: https://tracker.ceph.com/issues/42711 https://tracker.ceph.com/issues/42658 - pending backport https://github.com/ceph/ceph/pull/31461 - in progress, David is reviewing We need to get smoke in green state. Anything else for mimic? Thx YuriW On Thu, Nov 7, 2019 at 2:04 PM Casey Bodley <cbodley@redhat.com> wrote:
On 11/7/19 11:05 AM, Casey Bodley wrote:
On 11/7/19 10:47 AM, Yuri Weinstein wrote:
We need to add one more PR https://github.com/ceph/ceph/pull/31461
And also Casey is looking into surprising failures '!fails_on_rgw'" in rwg that have been introduced recently http://pulpito.ceph.com/yuriw-2019-11-07_01:09:25-rgw-wip-yuri-mimic_13.2.7_...
Some of the utf8 swift test cases just recently started failing, both here and in master. One of the errors is from httplib, saying 'InvalidURL: URL can't contain control characters.' So I suspect that httplib or some other python package changed and broke these tests. Is anyone aware of specific changes that could cause this?
It looks like this change in httplib was the result of a cve bug https://bugs.python.org/issue30458, where control characters in the url could be used to inject headers. I opened https://github.com/ceph/swift/pull/8 to disable the affected test cases, and will get that on the relevant release branches.
So false start, we will continue after these items are done.
On Wed, Nov 6, 2019 at 8:34 AM Yuri Weinstein <yweinste@redhat.com> wrote:
Hello
All PRs were tested and merged, mimic repo is locked and we will start QE validation soon.
Thx YuriW