These should only kick in at high debug levels. Moving any "work" between
the dout() and dendl lines puts them inside the magical hidden if block
that makes this code a no-op at low debug.
Signed-off-by: Sage Weil <sage@inktank.com>
We allow some trim during degraded, although we keep more entries around to
improve our chances of a restarting OSD of doing log-based recovery.
Still disallow during recovery...
Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Samuel Just <sam.just@inktank.com>
Chunky (and deep) scrub do not care about PG log trimming. Classic scrub
still does.
Deep scrub can take a long time, so not trimming the log during that period
may eat lots of RAM; avoid that!
Might fix: #4179
Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Samuel Just <sam.just@inktank.com>
With set -e this seems to fail (at least on some machines) if, say, there
is no MDS in the conf file. This fixes it.
Tested-by: Mark Nelson <mark.nelson@inktank.com>
Signed-off-by: Sage Weil <sage@inktank.com>
Small tweaks to the hadoop-internal test
to better use existing environment varaibles
and in response to the recent teuthology
changes.
Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Noah Watkins <noahwatkins@gmail.com>
Fix assert in replay(), check if mds->sessionmap.version is cmapv
instead of assign cmapv to mds->sessionmap.version. Add missing
second '='.
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Use empty() since it should be prefered as it has, following the
standard, a constant time complexity regardless of the containter
type. The same is not guaranteed for size().
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Use empty() since it should be prefered as it has, following the
standard, a constant time complexity regardless of the containter
type. The same is not guaranteed for size().
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Use empty() since it should be prefered as it has, following the
standard, a constant time complexity regardless of the containter
type. The same is not guaranteed for size().
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Use empty() since it should be prefered as it has, following the
standard, a constant time complexity regardless of the containter
type. The same is not guaranteed for size().
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Use empty() since it should be prefered as it has, following the
standard, a constant time complexity regardless of the containter
type. The same is not guaranteed for size().
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Use empty() since it should be prefered as it has, following the
standard, a constant time complexity regardless of the containter
type. The same is not guaranteed for size().
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Use empty() since it should be prefered as it has, following the
standard, a constant time complexity regardless of the containter
type. The same is not guaranteed for size().
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
The AC_PROG_CXX macro sets a flag if a C++ compiler is found
but does not fail if one is not found, it left to application
to test the flags as needed. This fix will issue an error
when a c++ compiler is not found. Bug 3955.
Signed-off-by: Gary Lowell <gary.lowell@inktank.com>
it's not installed, this fix adds an error message for a
qa/workunits/rbd/test_lock_fence.sh runs using test/rbdrw.py
rbdrw.py creates an image, locks it, and runs an I/O loop;
test_lock_fence.sh runs it, waits, and then blacklists that client,
which causes rbdrw.py to get ESHUTDOWN on operations thereafter.
Currently doesn't work with rbd caching enabled.
rbd.py gets new exception type for ESHUTDOWN
Fixes: #3190
Signed-off-by: Dan Mick <dan.mick@inktank.com>
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
We should use the PaxosServices getters, setters, and wait methods when and
wherever possible. These must have fallen through the cracks during the
merge.
Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
Only ignore errors after the midway point if the midway_sem_post is
defined.
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
(cherry picked from commit 5b24a68b6e)
The second test have pool deletion and object listing wait on the same
semaphore to connect and start. This led to errors sometimes when the
pool was deleted before it could be opened by the listing process. Add
another semaphore so the pool deletion happens only after the listing
has begun.
Fixes: #4147
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
(cherry picked from commit b0271e3905)
They need the same setup, and it's easy enough to run specific
subtests. Making them a separate subclass accidentally duplicated
tests from TestClone.
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
This error was masked before by watch notify not differentiating
between watches from the same client with different cookies.
Reopen the image at the end of this test so teardown works.
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
The second test have pool deletion and object listing wait on the same
semaphore to connect and start. This led to errors sometimes when the
pool was deleted before it could be opened by the listing process. Add
another semaphore so the pool deletion happens only after the listing
has begun.
Fixes: #4147
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
So we don't have to figure out which test is running from the output,
which can be difficult with the system tests.
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
Local async events are obsolete if the pg is deleting or if the
epoch at which they were created is prior to last_peering_reset.
Signed-off-by: Samuel Just <sam.just@inktank.com>
