Use OSD_POOL_PRIORITY_MAX and OSD_POOL_PRIORITY_MIN constants
Scale legacy priorities if exceeds maximum
Signed-off-by: David Zafman <dzafman@redhat.com>
Case 1: A more recent update exists
Case 2: The first entry in the divergent sequence is a create
Case 3 NOT TESTED - Ohject currently missing
Case 4: We can rollback all of the entries
Case 5: We cannot rollback at least 1 of the entries
Support starting OSDs even when "noup" is set (don't wait for up).
Move create_ec_pool() to ceph-helpers.sh
Fixes: https://tracker.ceph.com/issues/39162
Signed-off-by: David Zafman <dzafman@redhat.com>
stop command can be used to force stopping a specified osd daemon, e.g.,
you don't have to pre-figure out where it located.
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
The ceph cli tool checks for the presence of the variable, not its value.
Fixes: http://tracker.ceph.com/issues/38359
Signed-off-by: Sage Weil <sage@redhat.com>
Change run_osd() to default objectstore bluestore
Use run_osd_filestore() to use the non-default objectstore
Fix inject_eio to handle any objectstore if config prefixed with type
Remaining tests using filestore:
osd-pool-create.sh TEST_pool_create_rep_expected_num_objects
Test filestore directory creation
qa/standalone/osd/osd-dup.sh TEST_filestore_to_bluestore
Obvious
qa/standalone/osd/osd-rep-recov-eio.sh TEST_rep_read_unfound
Requires data digest in object info
qa/standalone/scrub/osd-scrub-repair.sh multiple tests
Erasure code pools append mode for filestore is tested
qa/standalone/special/ceph_objectstore_tool.py
Test code verifies COT by directly examining filestore contents
Fixes: https://tracker.ceph.com/issues/39162
Signed-off-by: David Zafman <dzafman@redhat.com>
* refs/pull/27169/head:
common/config: parse --default-$option as a default value
Reviewed-by: Sébastien Han <seb@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>
Sometimes it is useful to specify an alternative default value for an
option via the command line such that it has a lower priority than the
mon config database, config file, the rest of the command line, or the
environment.
Signed-off-by: Sage Weil <sage@redhat.com>
Leave repair pg state on until recovery finishes or a new scrub starts
Fixes: http://tracker.ceph.com/issues/38616
Signed-off-by: David Zafman <dzafman@redhat.com>
If the source or target PG version is 0'0, we may silently take the max
of the source and target and still leave the PG complete. This
specifically can happen with an empty PG, as seen with bug 38655. In
theory we could encounter one of the PGs with some other last_update
that doesn't match what we expect. If that ever happens, make sure the
result is incomplete so that backfill can clean up.
Additionally check that the pool metadata for the last merge matches the
PGs at all. This could mismatch if we have an osdmap gap and are forced
to do some merge without merge info at all... in which case we should
definitely invalidate: there should be newer copies of the PG(s), and we
have no idea whether the PGs we are merging are what we want. If this is
some disaster recovery situation, an operator is always free to use
ceph-objectstore-tool to re-mark a PG complete (at their own peril!).
Fixes: http://tracker.ceph.com/issues/38655
Signed-off-by: Sage Weil <sage@redhat.com>
- no need for the default pool size
- no initial osds or it will collide with setup_osds later
- no need for rbd pool at all
Signed-off-by: Sage Weil <sage@redhat.com>
'rbd pool init' now does IO. Drop the pool, or change the pool size to 1.
Fixes: http://tracker.ceph.com/issues/38585
Signed-off-by: Sage Weil <sage@redhat.com>
This test introduces a map gap. What *should* happen is that when there is
such a gap, we cannot import. Previously, the test didn't reliably produce
a map gap at all, and didn't check that import failed--it verified that it
passed.
Fix the test so that it reliably produces a gap *and* reports
min_last_epoch_clean to the mon so we can trim. Then verify we fail to
import, but can with --force. But remove the pg again, because if we
force an import with a map gap the osd will refuse to start.
Fixes: http://tracker.ceph.com/issues/38525
Signed-off-by: Sage Weil <sage@redhat.com>
The markdown test is based on marking down a specific number of times, but
the duplicate commands from the CLI may not get absorbed/batched by the
mon, breaking the test. Override the default qa/tasks/workunit.py
behavior of sending dups.
Fixes: http://tracker.ceph.com/issues/38359
Signed-off-by: Sage Weil <sage@redhat.com>
osd: Deny reservation if expected backfill size would put us over bac…
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>
* refs/pull/24828/head:
qa/osd-bluefs-volume-ops: use ceph-bluestore-tool for fsck
qa/osd-bluefs-volume-ops: reduce space usage for the test case
Reviewed-by: David Zafman <dzafman@redhat.com>
* refs/pull/24809/head:
os/bluestore: omit redundant '/' in OSD path for ceph-bluestore-tool if
os/bluestore: improve error handling for migrate ops in
qa/standtalone/osd-bluefs-volume-ops: remove redundant code.
Reviewed-by: Sage Weil <sage@redhat.com>
* refs/pull/24787/head:
Merge PR #24796 into nautilus
osd: fix heartbeat_reset unlock
Merge PR #24780 into nautilus
Merge PR #24761 into nautilus
Merge PR #24651 into nautilus
osd: fix race between op_wq and context_queue
test: Make sure kill_daemons failure will be easy to find
test: Add flush_pg_stats to make test more deterministic
* refs/pull/23985/head:
ceph-objectstore-tool: add back pool dne check
qa/suites/rados/singleton/reg11184: remove old test
ceph-objectstore-tool: import pg at original epoch
osd: handle null pg slot on startup
ceph-objectstore-tool: drop support for ancient export files
osd: avoid dropping osd_lock when pg osdmaps are not laggy
qa/standalone/osd/pg-merge.sh: add merge vs pg import test
Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
- In the jewel era, we fast-forwarded the PG to the OSD's latest epoch
and cleared past_intervals.
- In mimic, as of 2347ecb961, we brought the
PG up to date while updating past_intervals. (At the same time we removed
the OSD's parallel past_intervals regeneration.)
The problem is that the tool then has to reimplement the past_intervals
update logic, and *also* has to cope with splits and merges. Splits are
somewhat easier (until now we enable partial import of a PG into a split
child), but merges are not so easy.
This patch changes it so we import the PG and leave the pg_epoch matching
the import file. The OSD is then responsible for bringing it up to date
with the latest map, and dealing with any intervening splits or merges.
We also adjust the safety check to ensure that we don't collide with
any existing PG, either a child we eventually split into, or a parent
we eventually merge into.
Fixes: http://tracker.ceph.com/issues/35955
Signed-off-by: Sage Weil <sage@redhat.com>
- You can't import the source half a PG that's since merged. Sorry! We
could implement this later.
- You can import the target half, but the result will then be incomplete,
and you rely on backfill to clean it up.
- Map gaps don't affect this behavior.
Signed-off-by: Sage Weil <sage@redhat.com>
Grep from the primary's log, not every osd's log.
For the backfill_remapped task in particular, after the pg_temp change it
just so happens that the primary changes across the pool size change and
thus two different primaries do (some) backfill. Fix that test to pass
the correct primary.
Other tests are unaffected as they do not (happen to) trigger a primary
change and already satisfied the (removed) check that only one OSD does
backfill.
Signed-off-by: Sage Weil <sage@redhat.com>
Also:
- Do not print **offset** until specified
- Count missing objects correctly (used to be primary's local missing)
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
So if there are a lot fo missing objects on primary, we can
make use of auth_log_shard to restore client I/O quickly.
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
The log trimming case wasn't quite right. Before HEAD^ we were
rolling forward too aggressively and miscalculating the can_rollforward_to,
which affected the trim_to calculation.
Signed-off-by: Sage Weil <sage@redhat.com>
If the cluster dies during the rados bench, the maximum running time is
no more considered and all emitted aios are pending.
rados bench never quits and the global testing timeout (3600 sec : 1
hour) have to be reach to get a failure.
This situation is dramatic for a background test or a CI run as it locks
the whole job for too long for an event that will never occurs.
This ideal solution would be having 'rados bench' considering a failure
once the timeout is reached when aios are pending.
A possible workaround here is to put use the system command 'timeout'
before calling rados bench and fail if rados didn't completed on time.
To avoid side effects, this patch is doubling rados timeout. If rados
didn't completed after twice the expected time, it have to fail to avoid
locking the whole testing job.
Please find below the way it worked on a real test case.
We can see no IO after t>2 but despite timeout=4 the bench continue.
Thanks to this patch, the bench is stopped at t=8 and return 1.
5: /home/erwan/ceph/src/test/smoke.sh:55: TEST_multimon: timeout 8 rados -p foo bench 4 write -b 4096 --no-cleanup
5: hints = 1
5: Maintaining 16 concurrent writes of 4096 bytes to objects of size 4096 for up to 4 seconds or 0 objects
5: Object prefix: benchmark_data_mr-meeseeks_184960
5: sec Cur ops started finished avg MB/s cur MB/s last lat(s) avg lat(s)
5: 0 0 0 0 0 0 - 0
5: 1 16 1144 1128 4.40538 4.40625 0.00412965 0.0141116
5: 2 16 2147 2131 4.16134 3.91797 0.00985654 0.0109079
5: 3 16 2147 2131 2.77424 0 - 0.0109079
5: 4 16 2147 2131 2.0807 0 - 0.0109079
5: 5 16 2147 2131 1.66456 0 - 0.0109079
5: 6 16 2147 2131 1.38714 0 - 0.0109079
5: 7 16 2147 2131 1.18897 0 - 0.0109079
5: /home/erwan/ceph/src/test/smoke.sh:55: TEST_multimon: return 1
5: /home/erwan/ceph/src/test/smoke.sh:18: run: return 1
Signed-off-by: Erwan Velu <erwan@redhat.com>
Consolidate check() code and common script code
TEST_recovery_multi() wasn't reliable due to delayed peer_missing
Signed-off-by: David Zafman <dzafman@redhat.com>
This prevents the fix for http://tracker.ceph.com/issues/22050 or
potential future bugs from causing too much latency by trimming too
many log entries at once.
Signed-off-by: Josh Durgin <jdurgin@redhat.com>
Regular updates piggyback some osd state for this purpose with
MOSDRepOp[Reply]. Do the same thing for pure log entry updates (write
errors and lost/revert additions) via MOSDPGUpdateLogMissing[Reply].
Fixes: http://tracker.ceph.com/issues/22050
Signed-off-by: Josh Durgin <jdurgin@redhat.com>
Add test script that verifies the command in qa/standalone/osd
Fixes: http://tracker.ceph.com/issues/23242
Signed-off-by: Josh Durgin <jdurgin@redhat.com>
Signed-off-by: David Zafman <dzafman@redhat.com>
some tests, like osd-backfill-stats.sh are using delete_pool(), but
they don't have this function defined. and this function is defined
in standalone tests separately, so would be simpler if we can
consolidate them in ceph-helper.sh.
Signed-off-by: Kefu Chai <kchai@redhat.com>
osd/PG: re-write of _update_calc_stats and improve pg degraded state
Fixes: http://tracker.ceph.com/issues/20059
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Mon send osdmap to random osds after we mark osd down, the down osd
may use more than $sleep time to get updated osdmap if there is no
osd ping between osds. So create pool after setup cluster.
Signed-off-by: huangjun <huangjun@xsky.com>
wait_for_clean() can miss the new pool if it races with pool create.
Fixes: http://tracker.ceph.com/issues/20465
Signed-off-by: David Zafman <dzafman@redhat.com>
116cf759c8
will now hide all shadow trees(roots), so this is not applicable anymore
(actually it is misleading).
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
/bin/bash is a Linuxism. Other operating systems install bash to
different paths. Use /usr/bin/env in shebangs to find bash.
Signed-off-by: Alan Somers <asomers@gmail.com>
- stop running via make check
- add teuthology yamls to run them
- disable ceph_objecstore_tool.py for now (too slow for make check, and
we can't use vstart in teuthology via a package install)
- drop cephtool tests since those are already covered by other teuthology
tests
- leave a handful of (fast!) ceph-helpers tests for make check for minimal
integration tests.
Signed-off-by: Sage Weil <sage@redhat.com>
