* refs/pull/40326/head:
client: rebuild bl to avoid too many vector(> IOV_MAX)
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Otherwise pending_conns vector just keeps growing with redundant
connections all trying to reach the same set of monitors. When one
of the attempts finally succeeds, _finish_auth() will pick the first
connection with a matching entity_addr_t, designate it as active and
close all others. The match is very likely to be wrong and hence the
actual authenticated connection gets closed, leaving the OSD with
a bogus active_con and no monitor session.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
* refs/pull/40321/head:
cephadm: fix parsing of keepalived version (drop leading 'v')
cephadm: keepalived needs --cap-add=NET_RAW
cephadm: fix --cap-add=NET_ADMIN
cephadm: fix quoting for keepalived env var
mgr/cephadm: ha-rgw: use correct port
Reviewed-by: Michael Fritch <mfritch@suse.com>
* refs/pull/40015/head:
cephadm: validate fsid during cephadm shell command
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Juan Miguel Olmo <jolmomar@redhat.com>
GC behavior still needs a ton of work, but this adjustment should
reduce gc activity for now until we have time to refine the params
and introduce the ability to write cold extents to a segment other
than the current journal segment.
Signed-off-by: Samuel Just <sjust@redhat.com>
There's a performance overhead, but seastore is quite immature and
detecting these logic errors saves a ton of debugging time.
Signed-off-by: Samuel Just <sjust@redhat.com>
It's been almost three years and support is present in all relevant
clients.
From the security perspective, roughly the same could be achieved
with "ceph osd set-require-min-compat-client nautilus", but this is
more user friendly as the client gets ENOTSUP instead of spinning on
"feature set mismatch" faults.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Without this, `cephfs-top` hits an exception since the additional
metrics keys were not configured.
Also, include a validation suring selftest that checks if `ceph
fs perf stats` metrics match what cephfs-top is configured to
report.
Fixes: http://tracker.ceph.com/issues/49974
Signed-off-by: Venky Shankar <vshankar@redhat.com>
to enable XUnit plugin of jenkins to consume the ctest output and
publish it in the dashboard, we need to
* let ctest generate XML output instead of plain text output
* do not fail the test if any test case fails. this allows the publisher
to do its job by checking the XML output.
* prevent ctest from compressing the output. see
https://issues.jenkins.io/browse/JENKINS-21737
Signed-off-by: Kefu Chai <kchai@redhat.com>
run-promtool-unittests is failing with difference in floating point values in some complex calculations. This PR intends to simplify those calculations and fix this issue.
Fixes: https://tracker.ceph.com/issues/49952
Signed-off-by: Aashish Sharma <aasharma@redhat.com>
this change partially reverts b37e959238,
which introduced a regression where we fail to handle MOSDPGInfo
messages sent by osd before aba13c7661
and/or f7130db7a9fa85e475f14e4feed0097bf2d0b964.
Fixes: https://tracker.ceph.com/issues/49963
Signed-off-by: Kefu Chai <kchai@redhat.com>
so we can use the latest osdmap to tell what is the required osd release in
*current* cluster before consuming a new osdmap which is fed to
PeeringState only if we advance to a new map.
a osd shares its pg_info using MOSDPGInfo2 or MOSDPGInfo according to
PeeringState::last_require_osd_release, this happens before a new osdmap
is fully consumed.
Signed-off-by: Kefu Chai <kchai@redhat.com>
This makes
Mar 24 12:00:32 dael conmon[3969650]: Wed Mar 24 16:00:32 2021: cant open raw socket. errno=1
go away and allows it to enter the MASTER state.
Signed-off-by: Sage Weil <sage@newdream.net>
While any code that used the option is gone, it appears the option
itself was accidentally resurrected in various options.cc refactors
and transitions.
Signed-off-by: Adam C. Emerson <aemerson@redhat.com>
