History log of /4.6.4/ep-engine/ (Results 1 - 25 of 5706)
Revision (<<< Hide revision tags) (Show revision tags >>>)Date Author Comments
(<<< Hide modified files)
(Show modified files >>>)
Revision tags: v4.6.5, v4.6.3, v4.6.4
e6286bdc27-Apr-2017 Dave Rigby <daver@couchbase.com>

MB-24142: Use correct unit for slowTask recording

Change-Id: I388eb43fea01f8b79f5a122afa2c68757736fb81
Reviewed-on: http://review.couchbase.org/84424
Well-Formed: Build Bot <build@co

MB-24142: Use correct unit for slowTask recording

Change-Id: I388eb43fea01f8b79f5a122afa2c68757736fb81
Reviewed-on: http://review.couchbase.org/84424
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Daniel Owen <owend@couchbase.com>

show more ...

Revision tags: v4.6.2
4ea2003429-Mar-2017 Daniel Owen <owend@couchbase.com>

[BP] MB-25798: Don't schedule backfill until previous is complete

Previously in the handleSlowStream function if we are in the
Backfilling state and backfillTask is not running then we c

[BP] MB-25798: Don't schedule backfill until previous is complete

Previously in the handleSlowStream function if we are in the
Backfilling state and backfillTask is not running then we called
scheduleBackfill_UNLOCKED.

However although the backfillTask is not running there could still
be items in the Ready Queue.

As we are still in the Backfilling state, regardless as to whether
the backfillTask is running, we should drop the current cursor
and set the pendingBackfill flag.

Change-Id: I7d8c19041b9cec10640c4ef72c5d62cd73985ea4
Reviewed-on: http://review.couchbase.org/83533
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

2513928d08-Mar-2017 Daniel Owen <owend@couchbase.com>

[BP] MB-25798: Re-register a dropped cursor if we don't backfill

After dropping a cursor, when scheduleBackfill_UNLOCKED is called but
the backfill task does not need to be scheduled, it

[BP] MB-25798: Re-register a dropped cursor if we don't backfill

After dropping a cursor, when scheduleBackfill_UNLOCKED is called but
the backfill task does not need to be scheduled, it means the cursor is
not re-registered in markDiskSnapshot.

Therefore the cursor must therefore be re-registered from within
scheduleBackfill_UNLOCKED.

Change-Id: I4a643aed902316e0753555564b8bfd9b56291efe
Reviewed-on: http://review.couchbase.org/83531
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

Revision tags: v4.6.1
0286542c28-Feb-2017 Daniel Owen <owend@couchbase.com>

[BP] MB-25798: Backfill task leave stream state unchanged

The backfill task must not change the state machine of the active
stream that it is associated with. In particular:
1) In A

[BP] MB-25798: Backfill task leave stream state unchanged

The backfill task must not change the state machine of the active
stream that it is associated with. In particular:
1) In ActiveStream::markDiskSnapshot if a vbucket is not found then
it should just return, leaving the entity deleting the vbucket to
set the stream to dead.
2) ActiveStream::completeBackfill should not call
scheduleBackfill_UNLOCKED as this can move the stream into the
STREAM_IN_MEMORY state.

State machine changes should only be driven by ActiveStream::next,
which is invoked by DcpProducer::getNextItem.

The call to scheduleBackfill_UNLOCKED that was invoked in
ActiveStream::completeBackfill when the pendingBackfill flag was true
has been moved to the ActiveStream::backfillPhase function and is
invoked once the current backfill has completed and the pendingBackfill
flag is set to true.

Change-Id: Idcbf8164792f4fd09898fe90748424687c60fb6a
Reviewed-on: http://review.couchbase.org/83524
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

a2d57ae119-Sep-2017 Manu Dhundi <manu@couchbase.com>

[BP] MB-25798: Log when DcpProducer::Buffer log is full

Log when unable to notify a paused connection because the BufferLog is
full. And log when space becomes available in the BufferLo

[BP] MB-25798: Log when DcpProducer::Buffer log is full

Log when unable to notify a paused connection because the BufferLog is
full. And log when space becomes available in the BufferLog allowing a
paused connection to be notified.

Also log when ignoring a request to notify that a stream is ready due
to the associated vbucket already being in the ready queue.

Change-Id: I75f2d25af5d0f11175beb7b23300486092aa86fa
Reviewed-on: http://review.couchbase.org/83521
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

0b5d8a3c23-Feb-2017 Daniel Owen <owend@couchbase.com>

[BP] MB-25798: Log seqno data for when not scheduling backfill

When ActiveStream::scheduleBackfill_UNLOCKED is invoked with reschedule
defined to true it is rare for a backfill not to be

[BP] MB-25798: Log seqno data for when not scheduling backfill

When ActiveStream::scheduleBackfill_UNLOCKED is invoked with reschedule
defined to true it is rare for a backfill not to be scheduled.

In this case it is useful to log seqno data.

Change-Id: I145da4f69db3b81e77a4e85ff65ad070edf21c0e
Reviewed-on: http://review.couchbase.org/83520
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

33dac04321-Feb-2017 Daniel Owen <owend@couchbase.com>

[BP] MB-25798: Improved logging for handle slow stream & scheduling backfill

An example of the cause of the issue reported in MB-22451 is provided
below:

1/ DCP (Producer) eq_dc

[BP] MB-25798: Improved logging for handle slow stream & scheduling backfill

An example of the cause of the issue reported in MB-22451 is provided
below:

1/ DCP (Producer) eq_dcpq:replication:ns_1@172.23.96.101->ns_1@172.23.96.102:bucket-1 - (vb 399) Creating stream with start seqno 0 and end seqno 18446744073709551615
2/ DCP (Producer) eq_dcpq:replication:ns_1@172.23.96.101->ns_1@172.23.96.102:bucket-1 - (vb 399) Producer is handling slow stream; state:in-memory lastReadSeqno:548720 lastSentSeqno:538712
3/ DCP (Producer) eq_dcpq:replication:ns_1@172.23.96.101->ns_1@172.23.96.102:bucket-1 - (vb 399) Scheduling backfill from 548721 to 573764, reschedule flag : True
4/ DCP (Producer) eq_dcpq:replication:ns_1@172.23.96.101->ns_1@172.23.96.102:bucket-1 - (vb 399) Sending disk snapshot with start seqno 0 and end seqno 573764
5/ DCP (Producer) eq_dcpq:replication:ns_1@172.23.96.101->ns_1@172.23.96.102:bucket-1 - (vb 399) Producer is handling slow stream; state:backfilling lastReadSeqno:549860 lastSentSeqno:549140
6/ DCP (Producer) eq_dcpq:replication:ns_1@172.23.96.101->ns_1@172.23.96.102:bucket-1 - (vb 399) Backfill complete, 3989 items read from disk, 21055 from memory, last seqno read: 573764

A backfill is scheduled (step 3) and then before it completes, another
call is made to handle slow stream (step 5). This should result in
setting the pendingBackfill flag to true and then when the backfill
completes (step 6) it should schedule another backfill.
However this second backfill is not scheduled.

This patches adds improved logging which will provide the additional
information needed to debug this and similar issues.

Change-Id: I4a241c8f01c9dc828564649a5bdfd24be853308b
Reviewed-on: http://review.couchbase.org/83519
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

54e8ef6730-Aug-2017 Dave Rigby <daver@couchbase.com>

MB-25630: Include read-only KVStore stats in 'cbstats kvtimings'

During investigation of slow background fetches on a customer
environment, I found that while we record filesystem timing

MB-25630: Include read-only KVStore stats in 'cbstats kvtimings'

During investigation of slow background fetches on a customer
environment, I found that while we record filesystem timings for both
read-only and read-write KVStores, we only report the timings for
read-write.

To better assist in analysing read timings, we should also include the
read-only instance in the stats (accessed via the kvtimings cbstats
group).

Change-Id: Ie505e194cb1a91c80e8551cb198636ad68f56f1c
Reviewed-on: http://review.couchbase.org/82911
Tested-by: Build Bot <build@couchbase.com>
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: Manu Dhundi <manu@couchbase.com>

show more ...

d873ac7331-Jul-2017 Dave Rigby <daver@couchbase.com>

MB-25464: Do not exit cbstats if stats_perform fails

Exiting early will prevent cbstats collecting stats from other buckets
when running with `-a`.

Change-Id: I36641b06c5c22ea0a

MB-25464: Do not exit cbstats if stats_perform fails

Exiting early will prevent cbstats collecting stats from other buckets
when running with `-a`.

Change-Id: I36641b06c5c22ea0add46b3bbe91332ec0fc498a
Reviewed-on: http://review.couchbase.org/81627
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: James Harrison <james.harrison@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

0a68c87927-Jul-2017 Jim Walker <jim@couchbase.com>

MB-25238: Add logging to destruction of store

The MB shows lots of time missing between unregisterTaskable and
back in memcached when EvpDestroy has returned, this covers deletion
of

MB-25238: Add logging to destruction of store

The MB shows lots of time missing between unregisterTaskable and
back in memcached when EvpDestroy has returned, this covers deletion
of various objects and interestingly the destruction and flush of the
mutation log objects.

Change-Id: I3dd7913a5fb02e777b931a11eac97657584b79ee
Reviewed-on: http://review.couchbase.org/81481
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

73d8447221-Jul-2017 Jim Walker <jim@couchbase.com>

[BP] MB-25337: Handle the vbucket state change gracefully during rollback

Rollback is done asynchronously in kv-engine. When a scheduled rollback
task is run, the vbucket state might hav

[BP] MB-25337: Handle the vbucket state change gracefully during rollback

Rollback is done asynchronously in kv-engine. When a scheduled rollback
task is run, the vbucket state might have already changed to
non-replica. Upon such a condition, rollback task must just finish
running (as a null operation) rather than throwing an exception.

Change-Id: I459768be3727ca19e141a917e38892f91b5e43f9
Reviewed-on: http://review.couchbase.org/81238
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: Daniel Owen <owend@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

1d75037615-Jun-2017 Manu Dhundi <manu@couchbase.com>

MB-24817: During takeover, hold stream lock until vb is set to dead

During DCP takeover, when we decide to set the vbucket to dead state
on the producer side, we also transition the stre

MB-24817: During takeover, hold stream lock until vb is set to dead

During DCP takeover, when we decide to set the vbucket to dead state
on the producer side, we also transition the stream state from
STREAM_TAKEOVER_WAIT to STREAM_TAKEOVER_SEND state to send out any
remaining items that were received before the vbucket was set to
dead state.

The stream lock must be held until the vbucket is set to the dead
state, so that we do not prematurely finish sending out the last
items when the vbucket is not dead yet (that is if the vbucket
is not dead, it could get some more items).

This commit addresses the issue and also handles the ordering of
locks involved at stream level, ep store level and vbucket level
in the scenario.

Change-Id: I89bb42edec4f3765c8a9c67e6e89e9680eb40875
Reviewed-on: http://review.couchbase.org/79532
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

43b0f32e28-Apr-2017 Manu Dhundi <manu@couchbase.com>

MB-24148: Increase the backfill buf to avert hangs with large items

Currently we have a bug in backfill buffer accounting wherein
there could be a hang potentially with large items whose

MB-24148: Increase the backfill buf to avert hangs with large items

Currently we have a bug in backfill buffer accounting wherein
there could be a hang potentially with large items whose size are
almost equal to default max.

This commit addresses the bug by masking it. That is we increase the
default backfill buffer size to a value such that we cannot hit the
bug with default item max size.

We are just masking the bug to reduce the amount of change on this
maintainence branch. On the master we are fully fixing the bug.

Change-Id: I215c496cf92a9e9722291ac9a2ad6d14a0298f5c
Reviewed-on: http://review.couchbase.org/77474
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

879f176108-May-2017 Daniel Owen <owend@couchbase.com>

MB-24066/MB-22178: Set opencheckpointid to 1 after rollback

An opencheckpoint of 0 has the special meaning of being in a backfill
phase.

Therefore after performing a rollback we

MB-24066/MB-22178: Set opencheckpointid to 1 after rollback

An opencheckpoint of 0 has the special meaning of being in a backfill
phase.

Therefore after performing a rollback we must ensure the
opencheckpointid is reset to 1. This is the value used when the
vbucket is originally intialised.

Change-Id: I6b97faa7b502406961a4b48ad7affdbf6bc30512
Reviewed-on: http://review.couchbase.org/77837
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

9233046802-May-2017 Daniel Owen <owend@couchbase.com>

MB-24066: Partial revert "MB-22178: Don't use opencheckpointid to determine if in backfill phase"

Revert functional changes in
Change-Id: Ia977d6bf90e54fd1ceb8db4a9088b19d94d4bc8c,
w

MB-24066: Partial revert "MB-22178: Don't use opencheckpointid to determine if in backfill phase"

Revert functional changes in
Change-Id: Ia977d6bf90e54fd1ceb8db4a9088b19d94d4bc8c,
which although addressed the rollback bug described in MB-22178, caused
the bug described in MB-24066.

The tests added in Ia977d6bf90e54fd1ceb8db4a9088b19d94d4bc8c remain but
the rollback test has been disabled as it attempts a streamRequest when
the vbucket has a opendcheckpointid of 0.

The test will be re-enabled and modified with a follow-up patch that
addresses the bug described in MB-22178.

Change-Id: Ifd11c77a10e4ebe571c50e5d518403b423c3dfa5
Reviewed-on: http://review.couchbase.org/77604
Well-Formed: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

b22cf75903-Apr-2017 Jim Walker <jim@couchbase.com>

MB-23591: Re-instate isBucketDeletion check for flusher

The VBDeleteTask maybe scheduled to run, and isBucketDeletion
flag indicates if so, thus the flusher still needs to check
it t

MB-23591: Re-instate isBucketDeletion check for flusher

The VBDeleteTask maybe scheduled to run, and isBucketDeletion
flag indicates if so, thus the flusher still needs to check
it to ensure no writes occur to a file that is about to be
deleted.

Change-Id: I746eb372f35c9b838dfab2ee7ac2b6c59606e94a
Reviewed-on: http://review.couchbase.org/76186
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: David Haikney <david.haikney@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

7e406a4a23-Mar-2017 Manu Dhundi <manu@couchbase.com>

MB-23503: With an unpersisted snapshot, remove HT items till rollback point

When the rollback request intends to have a rollback to a point in
an unpersisted snapshot, we must remove all

MB-23503: With an unpersisted snapshot, remove HT items till rollback point

When the rollback request intends to have a rollback to a point in
an unpersisted snapshot, we must remove all unpersisted checkpoint
items from both checkpoint and hash table till the last persisted
disk snapshot.

Prior to this commit, we removed the items from the checkpoint
correctly, but only removed hash table items till the requested
rollback point, not all the unpersisted items, when
requested_rollback_seqno > persisted_seqno.

Note: We currently always rollback to a point which is persisted on
disk. Hence we must drop all checkpoint items from checkpoint and
hash table, irrespective of whether they are part of a full snapshot
or partial snapshot.

Change-Id: I1c14d0df6ae5e5459e60ecb0fc4a72ecd14231d3
Reviewed-on: http://review.couchbase.org/75628
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

ee2e2d4402-Mar-2017 Jim Walker <jim@couchbase.com>

MB-23112 Fix deleteWithMeta and extended meta data input

An incorrect offset meant that we tried to parse an incorrect
part of the packet as ExtendedMetaData.

Test code updated

MB-23112 Fix deleteWithMeta and extended meta data input

An incorrect offset meant that we tried to parse an incorrect
part of the packet as ExtendedMetaData.

Test code updated to cover this case.

Change-Id: If0610fe73b378e5830b4f4b9978d2b7507b516b4
Reviewed-on: http://review.couchbase.org/74529
Reviewed-by: David Haikney <david.haikney@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>
Well-Formed: Build Bot <build@couchbase.com>

show more ...

7b2c0f6319-Dec-2016 Daniel Owen <owend@couchbase.com>

MB-22180: Ensure all pendingBGFetches are terminated on VBbucket delete

On a VBucket delete we need to ensure that all pendingBGFetches are
terminated and any connections waiting for a B

MB-22180: Ensure all pendingBGFetches are terminated on VBbucket delete

On a VBucket delete we need to ensure that all pendingBGFetches are
terminated and any connections waiting for a BGFetch to complete are
notified with ENGINE_NOT_MY_VBUCKET.

Although we previously deleted the pendingBGFetches in the vbucket
destructor we did not send any notifications to waiting connections.

This patch moves the deletion of pendingBGFetches into the
notifyAllPendingConnsFailed function and in addition notifies all the
connections awaiting a BGFetch with an ENGINE_NOT_MY_VBUCKET.

Change-Id: I13a99fe01153a4ba8786aaf608b25ed31ace5a0c
Reviewed-on: http://review.couchbase.org/73811
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

17d8153f08-Dec-2016 Daniel Owen <owend@couchbase.com>

MB-22178: Don't use opencheckpointid to determine if in backfill phase

The opencheckpointid of a bucket can be zero after a rollback.
If an opencheckpointid was zero it was assumed that

MB-22178: Don't use opencheckpointid to determine if in backfill phase

The opencheckpointid of a bucket can be zero after a rollback.
If an opencheckpointid was zero it was assumed that the vbucket was in
backfilling state. This caused the producer stream request to be stuck
waiting for backfilling to complete.

Ths patch uses the vb->isBackfillPhase() to determine if the vbucket is
in a backfill state as opposed to using an opencheckpointid of zero.

Change-Id: Ia977d6bf90e54fd1ceb8db4a9088b19d94d4bc8c
Reviewed-on: http://review.couchbase.org/70810
Reviewed-by: Manu Dhundi <manu@couchbase.com>
Tested-by: buildbot <build@couchbase.com>
Reviewed-by: Jim Walker <jim@couchbase.com>
Reviewed-on: http://review.couchbase.org/73813
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>

show more ...

d7e1791616-Feb-2017 Daniel Owen <owend@couchbase.com>

MB-22256: Fix intermittent failures in test_duplicate_items_disk

The ep_testsuite: test_duplicate_items_disk test occassionally fails
because waiting for the ep_vbucket_del stat to updat

MB-22256: Fix intermittent failures in test_duplicate_items_disk

The ep_testsuite: test_duplicate_items_disk test occassionally fails
because waiting for the ep_vbucket_del stat to update (from 0 to 1)
times-out.

The reason the stat occasionally does not increase is because the
the vbucket that is undergoing deletion is set back to active before
the deletion is complete.

A side-effect of setting a vbucket to active, is that if the vbucket
does not exist then it is recreated.

This patch moves the setting of the vbucket back to active to after we
have confirmed the deletion has completed (by ensuring the
ep_vbucket_del stat increases). Thereby removing the race condition.

Change-Id: Ied680c64536862829974fd061a25d317f766bb46
Reviewed-on: http://review.couchbase.org/73758
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Jim Walker <jim@couchbase.com>
Reviewed-by: Manu Dhundi <manu@couchbase.com>

show more ...

370f70cd03-Feb-2017 Daniel Owen <owend@couchbase.com>

MB-22451: Ensure isBackfillTaskRunning is correctly set

In ActiveStream::completeBackfill if in STREAM_BACKFILLING state and
pendingBackfill is true then we will schedule another backfil

MB-22451: Ensure isBackfillTaskRunning is correctly set

In ActiveStream::completeBackfill if in STREAM_BACKFILLING state and
pendingBackfill is true then we will schedule another backfill. This
will cause isBackfillTaskRunning to be set to true. The flag should
remain true on exit of the completeBackfill function.

Change-Id: If8219a7f87b65af46d37a800eebf2257917cc555
Reviewed-on: http://review.couchbase.org/73310
Well-Formed: Build Bot <build@couchbase.com>
Tested-by: Build Bot <build@couchbase.com>
Reviewed-by: Jim Walker <jim@couchbase.com>

show more ...

2b2cf1be09-Jan-2017 Jim Walker <jim@couchbase.com>

MB-22235: Drift thresholds don't get updated from set_param

Need to plug the change value listener into the bucket config
object.

Test updated to look at the vbucket's threshold

MB-22235: Drift thresholds don't get updated from set_param

Need to plug the change value listener into the bucket config
object.

Test updated to look at the vbucket's threshold values

Change-Id: Id515d71fe551474e93ff5113148ff8411ed739fb
Reviewed-on: http://review.couchbase.org/72100
Well-Formed: buildbot <build@couchbase.com>
Tested-by: buildbot <build@couchbase.com>
Reviewed-by: Dave Rigby <daver@couchbase.com>

show more ...

Revision tags: v4.6.0
e4c8bcbb07-Dec-2016 Dave Rigby <daver@couchbase.com>

MB-21925: Fix queue_fill stat when persitenceCursor on queue_op::empty

There is a bug related to how we account for items when we enqueue
mutations to be persisted (Checkpoint::queueDirt

MB-21925: Fix queue_fill stat when persitenceCursor on queue_op::empty

There is a bug related to how we account for items when we enqueue
mutations to be persisted (Checkpoint::queueDirty). The bug occurs
when we enqueue an item which already exists (i.e. it is to be
de-duplicated), and the persistence cursor has not permitted any items
for that VBucket - i.e. at the very beginning of the lifetime of the
VBucket.

With the advent of the MB-20852 (the set_vbucket_state changes), the
first item in a Bucket's checkpoint is now a
queue_op::set_vbucket_state - previously it would have been a "normal"
item. When items are added to a checkpoint, we just update the queuing
stats - essentially there are two possibilities:

1. The item is a new key (NEW_KEY), or it's a key which isn't new but
has already been persisted (PERSIST_AGAIN) - we need to persist it -
increment queue_fill

2. The item can be de-duplcated (EXISTING_ITEM) - don't increment
queue_fill.

Essentially, there's a bug in this logic - we incorrectly mark the
incoming item as PERSIST_AGAIN, instead of EXISTING_ITEM. As a result
the queue_fill value is incorrectly inflated.

This is due to using a unsigned value for the mutation_id, which we
decrement by one if it's a meta_item. This should go to -1 (less than
any possible valid item id), but instead it goes to 2^63-1, *greater*
than any possible valid item (!)

Note: this is purely a stats issue, the actual item /is/ queued and
persisted correctly.

Change-Id: I8d40c417dc165e192b4f90e8cc16c6e87557d451
Reviewed-on: http://review.couchbase.org/70730
Tested-by: buildbot <build@couchbase.com>
Well-Formed: buildbot <build@couchbase.com>
Reviewed-by: Daniel Owen <owend@couchbase.com>

show more ...

d85969d501-Dec-2016 Dave Rigby <daver@couchbase.com>

MB-21867: test_item_pager: Retry on ENOMEM when fetching items

Change-Id: I9031ca1ff51aed609144ccf141897354138c2f88
Reviewed-on: http://review.couchbase.org/70515
Tested-by: buildbot

MB-21867: test_item_pager: Retry on ENOMEM when fetching items

Change-Id: I9031ca1ff51aed609144ccf141897354138c2f88
Reviewed-on: http://review.couchbase.org/70515
Tested-by: buildbot <build@couchbase.com>
Reviewed-by: Daniel Owen <owend@couchbase.com>

show more ...

12345678910>>...229