History log of /4.5.1-MP3/ns_server/ (Results 1 - 25 of 11187)
Revision (<<< Hide revision tags) (Show revision tags >>>)Date Author Comments
(<<< Hide modified files)
(Show modified files >>>)
Revision tags: 6.0.3, 5.5.5, 5.5.4, 6.0.1, 5.5.3, 6.0.0, 5.5.2, 5.1.2, 5.5.1, 5.5.0, 4.6.5, 5.1.0, 5.0.1, 4.6.4, 5.0.0, 4.6.3, 4.6.2, 4.6.0
ba2ecedd20-Jul-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

Merge remote-tracking branch 'couchbase/sherlock' into watson

* couchbase/sherlock:
MB-20162 Scale fragmentation size limit for regular vbuckets.
MB-20162 Short-circuit compactio

Merge remote-tracking branch 'couchbase/sherlock' into watson

* couchbase/sherlock:
MB-20162 Scale fragmentation size limit for regular vbuckets.
MB-20162 Short-circuit compaction if there are no vbuckets.
MB-20162 Ignore size threshold when compacting master db.
MB-20162 Don't normalize fragmentation until check_fragmentation.
MB-20102 Filter out dev ddocs correctly in get_safe_purge_seqs.

Conflicts:
src/compaction_new_daemon.erl

Change-Id: Iedb1063dab795b473dba57ff539350fd8c63910c

show more ...


Revision tags: 4.1.2
aaba603d14-Jul-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

MB-20162 Scale fragmentation size limit for regular vbuckets.

Change-Id: I22a9507520b022fe3eabb6d91b5bc51f4de81eeb
Reviewed-on: http://review.couchbase.org/65785
Tested-by: Aliaksey

MB-20162 Scale fragmentation size limit for regular vbuckets.

Change-Id: I22a9507520b022fe3eabb6d91b5bc51f4de81eeb
Reviewed-on: http://review.couchbase.org/65785
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Well-Formed: buildbot <build@couchbase.com>
Reviewed-by: Dave Finlay <dave.finlay@couchbase.com>

show more ...

64e6f85419-Jul-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

MB-20162 Short-circuit compaction if there are no vbuckets.

Change-Id: I793805c9d0def90f4ebf4283e6aec9600eaf9264
Reviewed-on: http://review.couchbase.org/65961
Tested-by: Aliaksey Ar

MB-20162 Short-circuit compaction if there are no vbuckets.

Change-Id: I793805c9d0def90f4ebf4283e6aec9600eaf9264
Reviewed-on: http://review.couchbase.org/65961
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Well-Formed: buildbot <build@couchbase.com>
Reviewed-by: Dave Finlay <dave.finlay@couchbase.com>

show more ...

2093364613-Jul-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

MB-20162 Ignore size threshold when compacting master db.

Fragmentation size threshold is a value that is supposed to be mean
the max absolute wasted space per bucket per node, so it doe

MB-20162 Ignore size threshold when compacting master db.

Fragmentation size threshold is a value that is supposed to be mean
the max absolute wasted space per bucket per node, so it doesn't make
sense on a per vbucket level (unless scaled properly).

And given that master vbuckets are quite special, we actually cheat and
ignore it altogether.

Change-Id: I36cdbace91903456a9f68e64788f484b871af59d
Reviewed-on: http://review.couchbase.org/65784
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Reviewed-by: Dave Finlay <dave.finlay@couchbase.com>
Well-Formed: buildbot <build@couchbase.com>

show more ...

56aaae0b13-Jul-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

MB-20162 Don't normalize fragmentation until check_fragmentation.

That allows upper level functions to make decisions based on the
values set by the user.

Change-Id: I58d677c7bd

MB-20162 Don't normalize fragmentation until check_fragmentation.

That allows upper level functions to make decisions based on the
values set by the user.

Change-Id: I58d677c7bdc5b17e0cc5bf64b1b21199543f5250
Reviewed-on: http://review.couchbase.org/65783
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Reviewed-by: Dave Finlay <dave.finlay@couchbase.com>
Well-Formed: buildbot <build@couchbase.com>

show more ...

Revision tags: 4.5.1
36f7074609-Jun-2016 Dave Rigby <daver@couchbase.com>

MB-18426: Reduce compaction_number_of_kv_workers from 4 to 1

There have been a number of instances where the IO cost of compaction
has had a detremental effect on normal "front-end" KV o

MB-18426: Reduce compaction_number_of_kv_workers from 4 to 1

There have been a number of instances where the IO cost of compaction
has had a detremental effect on normal "front-end" KV operations - see
MB-18426 for full details, but as one example:

In a heavy write use case where a document is written to once, read
once and then deleted. The documents are only stored in the bucket
until they are processed, so in a normal state a document only
exists for a few minutes. It was noted that when the 3 (see note)
compactors ran the disk would become fully saturated and this would
cause huge spikes in the disk queues up to 36 million. When the
compaction_number_of_kv_workers was reduced, it had the following
impacts:
* Reduce the disk queue to ten of thousands from 36 million,
which also reduce the amount of memory the bucket was using.
* Reduce the disk IO.
* Most surprisingly reducing the disk space required from 280GB to
40GB. I suspect this was because the deletes were being blocked on
the disk write queue.
* The fragmentation percentages fluctuates a lot more.

Similar results have been reported in 2 other customer enviroments.

This patch reduces the default number down to 1 - it can still be
overridden if necessary via diag/eval.

Note: While the value of compaction_number_of_kv_workers was
previously 4, we would normally only see 3 compactors run in
parallel in ep-engine due to not scheduling more than num_shards/2 +
1 compactors at once - see EventuallyPersistentStore::compactDB.

Change-Id: I87bc03681f93f1c7df5535181c7fa1c640cb8e68
Reviewed-on: http://review.couchbase.org/64850
Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
Reviewed-by: Artem Stemkovski <artem@couchbase.com>
Well-Formed: buildbot <build@couchbase.com>
Tested-by: Dave Rigby <daver@couchbase.com>

show more ...

94e6825907-Jul-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

MB-20102 Filter out dev ddocs correctly in get_safe_purge_seqs.

As of before, the code would just filter out desgin documents with one
character after _design/dev_ prefix.

The c

MB-20102 Filter out dev ddocs correctly in get_safe_purge_seqs.

As of before, the code would just filter out desgin documents with one
character after _design/dev_ prefix.

The consequence of this is that in case there's only development
design documents for some bucket, those won't get auto-updated. So the
compaction daemon would be passing ever more out of date purge seqno
to ep-engine, preventing it from reclaiming disk space.

Change-Id: I01dc63813373b0e466931091984f05591f80a523
Reviewed-on: http://review.couchbase.org/65592
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Reviewed-by: Dave Finlay <dave.finlay@couchbase.com>
Well-Formed: buildbot <build@couchbase.com>

show more ...

31b3c7ee28-Jun-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

Revert "Stop rebalance if memcached set cluster config fails"

This reverts commit 33b62a9d5a23e2f1903697a8ab3bbca3071ebe08.

Conflicts:
src/terse_bucket_info_uploader.er

Revert "Stop rebalance if memcached set cluster config fails"

This reverts commit 33b62a9d5a23e2f1903697a8ab3bbca3071ebe08.

Conflicts:
src/terse_bucket_info_uploader.erl

Change-Id: I7f2a68ad4aad2b06b2b25bf56f6c699ccf26fd7f
Reviewed-on: http://review.couchbase.org/65282
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

8b5b418a21-Jun-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

Merge remote-tracking branch 'couchbase/master' into watson

* couchbase/master: (32 commits)
Flush bucket_info_cache_invalidations to avoid unneeded cluster updates
Stop rebalanc

Merge remote-tracking branch 'couchbase/master' into watson

* couchbase/master: (32 commits)
Flush bucket_info_cache_invalidations to avoid unneeded cluster updates
Stop rebalance if memcached set cluster config fails
MB-19880: Add cbcollect_info output to detect containers
MB-19752 do not remove read-only user account when setting admin user
MB-19832 add 1 second delay before waking up if dcp_notifier returned error
MB-19832 reset last_known_pos in dcp_notifier if stream ends not with ok
MB-19608 pull config from ns_server node before starting capi https server
MB-19818: do not show failover warnings during rebalance
[rebuild_replicas] Sync ns_config before rebalance/bucket update.
[rebuild_replicas] Sleep for 10 seconds after the rebalances.
MB-19319 Dump cbcollect_info output if collection fails.
MB-19319 Reimplement stdin watcher using low-level IO.
MB-19706 Shutdown replica builders at the end of TAP moves.
Add a script to convert timestamps to human readable datetimes.
Add a script to extract cpu stats from stats archives.
Couple of jq scripts to work with master_events.
Die on extra data on socket after memcached call.
Keep last call processed by ns_memcached worker in process dict.
Keep socket name in ns_memcached workers dictionaries.
Use ns_config:latest() in master_activity_events where possible.
...

Change-Id: I5d6c89ffd3bd3fb379844c98a02c627d92365f28

show more ...


Revision tags: 3.1.6, 4.5.0
3a7f690706-Apr-2016 Anders Nygren <anders.nygren@couchbase.com>

Flush bucket_info_cache_invalidations to avoid unneeded cluster updates

terse_bucket_info_uploader is converted to a gen_server, and flushes
bucket_info_cache_invalidation messages in or

Flush bucket_info_cache_invalidations to avoid unneeded cluster updates

terse_bucket_info_uploader is converted to a gen_server, and flushes
bucket_info_cache_invalidation messages in order to avoid repeted
memcached cluster configuration updates with identical data.

Signed-off-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
[aliaksiej.artamonau@gmail.com: removed generic gen_server comments]

Change-Id: I935600d5ed098f93028bf46b754a63c2b9acb91e
Reviewed-on: http://review.couchbase.org/62505
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

33b62a9d05-Apr-2016 Anders Nygren <anders.nygren@couchbase.com>

Stop rebalance if memcached set cluster config fails

Change-Id: Id70f847eed1c7c25a0ff53f8d9012634aef461fc
Reviewed-on: http://review.couchbase.org/62467
Reviewed-by: Aliaksey Artamon

Stop rebalance if memcached set cluster config fails

Change-Id: Id70f847eed1c7c25a0ff53f8d9012634aef461fc
Reviewed-on: http://review.couchbase.org/62467
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

69f4603914-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

Merge remote-tracking branch 'gerrit/sherlock'

* MB-19752 do not remove read-only user account when setting admin user

Change-Id: If4d7a8f0063849936f5b09b85bbf7233f66ab672


dfebaced14-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

Merge remote-tracking branch 'gerrit/rel-3.0.0++'

* MB-19752 do not remove read-only user account when setting admin user

Change-Id: Iff89c743f4486a4e036df43dac4258e81c237233


ad9c1c7410-Jun-2016 Anders Nygren <anders.nygren@couchbase.com>

MB-19880: Add cbcollect_info output to detect containers

Change-Id: I33bb89a90da2ac686f28f79dbb5d3cfe67a86c62
Reviewed-on: http://review.couchbase.org/64889
Tested-by: Anders Nygren

MB-19880: Add cbcollect_info output to detect containers

Change-Id: I33bb89a90da2ac686f28f79dbb5d3cfe67a86c62
Reviewed-on: http://review.couchbase.org/64889
Tested-by: Anders Nygren <anders.nygren@couchbase.com>
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

0a707c2a07-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

MB-19752 do not remove read-only user account when setting admin user

I decided not to check if the admin and roadmin user names match because:

1. We won't be able to do it for LDAP

MB-19752 do not remove read-only user account when setting admin user

I decided not to check if the admin and roadmin user names match because:

1. We won't be able to do it for LDAP user
2. During the login the admin user is checked first, so whoever comes
with this common name will get full admin access. Which makes this
situation not harmful and easily correctable.

Change-Id: Ibf7d21146904edfffbe73286c1322da960bd618b
Reviewed-on: http://review.couchbase.org/64799
Well-Formed: buildbot <build@couchbase.com>
Tested-by: Artem Stemkovski <artem@couchbase.com>
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

d12b499d06-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

Merge remote-tracking branch 'gerrit/sherlock'

* MB-19608 pull config from ns_server node before starting capi https server
* MB-19832 add 1 second delay before waking up if dcp_notifier

Merge remote-tracking branch 'gerrit/sherlock'

* MB-19608 pull config from ns_server node before starting capi https server
* MB-19832 add 1 second delay before waking up if dcp_notifier returned error
* MB-19832 reset last_known_pos in dcp_notifier if stream ends not with ok

Change-Id: I585368c100ab8b3aad8a7cec31891d863049204d

show more ...


48c442bd06-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

Merge remote-tracking branch 'gerrit/rel-3.0.0++'

* MB-19832 reset last_known_pos in dcp_notifier if stream ends not with ok
* MB-19832 add 1 second delay before waking up if dcp_notifie

Merge remote-tracking branch 'gerrit/rel-3.0.0++'

* MB-19832 reset last_known_pos in dcp_notifier if stream ends not with ok
* MB-19832 add 1 second delay before waking up if dcp_notifier returned error

Change-Id: Ic24be2f8dc54c2994b84ccf04aeb552a1b18057d

show more ...


b5c06aa103-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

MB-19832 add 1 second delay before waking up if dcp_notifier returned error

this is needed to wait for things to settle out and prevent opening
unneeded streams. for example if the vbuck

MB-19832 add 1 second delay before waking up if dcp_notifier returned error

this is needed to wait for things to settle out and prevent opening
unneeded streams. for example if the vbucket changes status from active to
replica, the stream will end with status=2, and xdc_vbucket_rep will wake up,
won’t get any data and then open new notifier stream. some time later the
xdc_vbucket_rep will die, but stream will remain open, until status changes
again

1 sec delay will prevent this from happening in most of the cases

Change-Id: Iad4aade149691d1de7f3304ce4291879f65c25d9
Reviewed-on: http://review.couchbase.org/64700
Tested-by: Artem Stemkovski <artem@couchbase.com>
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Well-Formed: buildbot <build@couchbase.com>

show more ...

397638ea03-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

MB-19832 reset last_known_pos in dcp_notifier if stream ends not with ok

so next time when notifier will be asked about changes, the real stream
will be opened no matter what

Ch

MB-19832 reset last_known_pos in dcp_notifier if stream ends not with ok

so next time when notifier will be asked about changes, the real stream
will be opened no matter what

Change-Id: Iae1f75b94822b276872b0bcc1ccd9d7a3cc10673
Reviewed-on: http://review.couchbase.org/64699
Tested-by: Artem Stemkovski <artem@couchbase.com>
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Well-Formed: buildbot <build@couchbase.com>

show more ...

102c9f1706-Jun-2016 Artem Stemkovski <astemkov@gmail.com>

MB-19608 pull config from ns_server node before starting capi https server

to eliminate a race during capi https server restart, when config changes
might be delivered to ns_couchdb node

MB-19608 pull config from ns_server node before starting capi https server

to eliminate a race during capi https server restart, when config changes
might be delivered to ns_couchdb node only after https server is restarted,
causing capi https server to ignore changed configuration

Change-Id: I0b8714c998ffc19b4d7d03a46b83901f91f918f2
Reviewed-on: http://review.couchbase.org/64762
Well-Formed: buildbot <build@couchbase.com>
Tested-by: Artem Stemkovski <artem@couchbase.com>
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

5f03809f03-Jun-2016 Pavel Blagodov <stochmail@gmail.com>

MB-19818: do not show failover warnings during rebalance

Change-Id: Iee505cadc08bcb3869b82010264e1848d28c2968
Reviewed-on: http://review.couchbase.org/64715
Tested-by: Pavel Blagodov

MB-19818: do not show failover warnings during rebalance

Change-Id: Iee505cadc08bcb3869b82010264e1848d28c2968
Reviewed-on: http://review.couchbase.org/64715
Tested-by: Pavel Blagodov <stochmail@gmail.com>
Reviewed-by: Dave Finlay <dave.finlay@couchbase.com>

show more ...

9c0ca9ad31-May-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

Merge remote-tracking branch 'couchbase/sherlock'

* couchbase/sherlock:
[rebuild_replicas] Sync ns_config before rebalance/bucket update.
[rebuild_replicas] Sleep for 10 seconds

Merge remote-tracking branch 'couchbase/sherlock'

* couchbase/sherlock:
[rebuild_replicas] Sync ns_config before rebalance/bucket update.
[rebuild_replicas] Sleep for 10 seconds after the rebalances.
MB-19319 Dump cbcollect_info output if collection fails.
MB-19319 Reimplement stdin watcher using low-level IO.
MB-19706 Shutdown replica builders at the end of TAP moves.

Change-Id: I6a304bc4ac1ce3db3af8fe7d40583cdd1c293aab

show more ...


4f75617531-May-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

Merge remote-tracking branch 'couchbase/rel-3.0.0++' into sherlock

* couchbase/rel-3.0.0++:
[rebuild_replicas] Sync ns_config before rebalance/bucket update.
[rebuild_replicas] S

Merge remote-tracking branch 'couchbase/rel-3.0.0++' into sherlock

* couchbase/rel-3.0.0++:
[rebuild_replicas] Sync ns_config before rebalance/bucket update.
[rebuild_replicas] Sleep for 10 seconds after the rebalances.
MB-19319 Dump cbcollect_info output if collection fails.
MB-19319 Reimplement stdin watcher using low-level IO.
MB-19706 Shutdown replica builders at the end of TAP moves.

Change-Id: I9a9d528ea6331097bb5965cc8bcfa5c92c30936a

show more ...


3ea4b16727-May-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

[rebuild_replicas] Sync ns_config before rebalance/bucket update.

That minimizes the chances of conflicts even if mastership moves to a
different node.

Another source of config

[rebuild_replicas] Sync ns_config before rebalance/bucket update.

That minimizes the chances of conflicts even if mastership moves to a
different node.

Another source of config conflicts is the fact, that repl_type is
updated on the node executing the script, not the master. So this is
also addressed.

Change-Id: I696daf59c60e21934c385e9d7251ef9b7f3f888c
Reviewed-on: http://review.couchbase.org/64546
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

9562898427-May-2016 Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

[rebuild_replicas] Sleep for 10 seconds after the rebalances.

This is to prevent mb_master from restarting if the server doesn't have
this fix:: http://review.couchbase.org/64083.

[rebuild_replicas] Sleep for 10 seconds after the rebalances.

This is to prevent mb_master from restarting if the server doesn't have
this fix:: http://review.couchbase.org/64083.

mb_master_sup's restart policy is {one_for_one, 3, 10}, so just sleeping
for 10 seconds after two ns_orchestrator crashes should prevent
mb_master_sup from propagating the crash up to mb_master itself.

Change-Id: I5649cc2d168a9796c3a4f8e30f7d34e466a22342
Reviewed-on: http://review.couchbase.org/64545
Tested-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>
Reviewed-by: Aliaksey Artamonau <aliaksiej.artamonau@gmail.com>

show more ...

12345678910>>...448