History log of /6.0.3/forestdb/ (Results 251 - 275 of 1302)
Revision (<<< Hide revision tags) (Show revision tags >>>)Date Author Comments
(<<< Hide modified files)
(Show modified files >>>)
2752c9f128-Dec-2015 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Add an AVL-tree for reusable blocks

- Instead of linear scanning, we can quickly find next reusable block
by traversing the bitmap index (i.e., AVL-tree).

Change-Id: I035ebabff2

Add an AVL-tree for reusable blocks

- Instead of linear scanning, we can quickly find next reusable block
by traversing the bitmap index (i.e., AVL-tree).

Change-Id: I035ebabff290e306e56fd5cbe22ffad2ae5da1a5

show more ...

f0b1bf7707-Dec-2015 Jung-Sang Ahn <jungsang.ahn@gmail.com>

MB-16219 Support superblock and circular block reusing

- Super blocks point to the up-to-date DB header.

- When configured conditions are satisfied, stale blocks are reused
in a

MB-16219 Support superblock and circular block reusing

- Super blocks point to the up-to-date DB header.

- When configured conditions are satisfied, stale blocks are reused
in a circular manner. This does not increase the DB file size, so
that we can largely reduce the overhead from compaction.

- The latest a few old versions are preserved for future snapshot
creation.

Change-Id: I6eb9209ced436c0a84ebb0216344eb04f300fca9

show more ...

2210627a30-Jan-2016 Chiyoung Seo <chiyoung@couchbase.com>

Merge "CMake: Only build each source file once where possible - 2x build speedup"


03a8608c29-Jan-2016 Sundar Sridharan <sundar.sridharan@gmail.com>

init filemgr_config in filemgr_destroy_file, fix crash

Change-Id: Ied4d62e54c474906cb3b9c6ac39bf21f7e196e20

5c9158b029-Jan-2016 Dave Rigby <daver@couchbase.com>

CMake: Only build each source file once where possible - 2x build speedup

Use CMake OBJECT libraries (a collection of objects which are built
just once) for source files which are used i

CMake: Only build each source file once where possible - 2x build speedup

Use CMake OBJECT libraries (a collection of objects which are built
just once) for source files which are used identically in different
targets. This signifcantly reduces the number of object files built
(previoulsy CMake would build each source file once for each target is
is included in).

Significantly reduces build time - time for a clean build of the
forestdb/ subdir (ccache disabled) - make -j8

Object files: 628

Times:
real 0m50.530s
user 0m27.857s
sys 0m18.651s

Object files: 284

Times:
real 0m35.181s
user 0m21.727s
sys 0m11.485s

Change-Id: I13fcd339e126095ea5e58b9305c9864742bd5c8a

show more ...

c7b169cb28-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix memory leak in compaction_cancellation_test

- memleak_start() & memleak_end() are called twice in the same test.

Change-Id: Id1b77315b33cdd77559680c762be8fe572dace90

9ecf675525-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

Merge remote-tracking branch 'origin/sherlock'

* origin/sherlock:
MB-17009 use fprintf to log errors in a daemon compaction thread

Conflicts:
src/compactor.cc

Merge remote-tracking branch 'origin/sherlock'

* origin/sherlock:
MB-17009 use fprintf to log errors in a daemon compaction thread

Conflicts:
src/compactor.cc

Change-Id: I502f526712bad11216c6cf874bc2b220fcd91630

show more ...


7c9c39cb24-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

MB-17009 use fprintf to log errors in a daemon compaction thread

We identified that ForestDB daemon thread sometimes gets stuck inside
the log callback function provided by Go-based appl

MB-17009 use fprintf to log errors in a daemon compaction thread

We identified that ForestDB daemon thread sometimes gets stuck inside
the log callback function provided by Go-based application.

Until we figure out the root cause of this issue (i.e., c -> cgo-> go
call stack issue), we use fprintf to log any errors that can happen
inside ForestDB daemon compaction thread.

Change-Id: Icbf6d3440a295443b7476827329ef8f242b5a44c

show more ...

526efd9522-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

Merge remote-tracking branch 'origin/sherlock'

* origin/sherlock:
Revert "MB-17004 Don't flush the WAL for each commit in fdb_compact_upto"

Conflicts:
src/filemgr.

Merge remote-tracking branch 'origin/sherlock'

* origin/sherlock:
Revert "MB-17004 Don't flush the WAL for each commit in fdb_compact_upto"

Conflicts:
src/filemgr.cc
src/forestdb.cc

Change-Id: I4772a4f8e8f2c4c292843e8ec5154de920939d47

show more ...


d7da1b3922-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

Revert "MB-17004 Don't flush the WAL for each commit in fdb_compact_upto"

This reverts commit f3e4c58fd3e651eca09f98aaee2599d8cba4963e.

e57e33d316-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

Merge remote-tracking branch 'origin/sherlock'

* origin/sherlock:
MB-17004 Don't flush the WAL for each commit in fdb_compact_upto

Conflicts:
src/forestdb.cc

Merge remote-tracking branch 'origin/sherlock'

* origin/sherlock:
MB-17004 Don't flush the WAL for each commit in fdb_compact_upto

Conflicts:
src/forestdb.cc

Change-Id: I0a1bb47482ea0cf93a05ccecfe3ddfb96d38e849

show more ...


9e1a467011-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

MB-16947 Support fdb_cancel_compact API

This new API cancels the compaction task for a given file if it is
currently running.

Change-Id: Ia496ac8cfe15f5c8f63d88603c6f29e6822d71c6

f3e4c58f07-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

MB-17004 Don't flush the WAL for each commit in fdb_compact_upto

We don't need to flush the WAL for each commit while executing
fdb_compact_upto, but instead perform the WAL flush iff an

MB-17004 Don't flush the WAL for each commit in fdb_compact_upto

We don't need to flush the WAL for each commit while executing
fdb_compact_upto, but instead perform the WAL flush iff any of
WAL flush thresholds (memory or number of WAL items) is satisfied.

Change-Id: Ib52f4df0000b0b48d610359792eff1084a848fbb

show more ...

0c3fbf7d11-Jan-2016 Chiyoung Seo <chiyoung@couchbase.com>

Merge "use percentage value as regression threshold"


42b318ce11-Jan-2016 Trond Norbye <trond.norbye@gmail.com>

FreeBSD may use CLOCK_MONOTONIC

Change-Id: If89e22bde25d0f6fb02237d9716ff7ee5070f314

1904677708-Jan-2016 Sundar Sridharan <sundar.sridharan@gmail.com>

save 16 bytes per WAL item by list_elem_txn

wal item can't be in transaction list and flush tree
at same time. Putting it in the union to save 16 bytes per
WAL item which translates

save 16 bytes per WAL item by list_elem_txn

wal item can't be in transaction list and flush tree
at same time. Putting it in the union to save 16 bytes per
WAL item which translates to a big gain during compaction &
with large wal sizes

Change-Id: Ifb419507e6d8b0651c82a1f85b1fc371a18d73f8

show more ...

8122b56d06-Jan-2016 Sundar Sridharan <sundar.sridharan@gmail.com>

MB-17215: Add log_callbacks, new testfilenames fdb_extended_test

Not having log_callbacks defined was resulting in errors like
+ log_callback 0x00000026866ed4e8
{callback=0xcccc

MB-17215: Add log_callbacks, new testfilenames fdb_extended_test

Not having log_callbacks defined was resulting in errors like
+ log_callback 0x00000026866ed4e8
{callback=0xcccccccccccccccc ctx_data=0xcccccccccccccccc
}
err_log_callback *
log_callback->callback 0xcccccccccccccccc void
(int, const char *,
void *) *
log_callback->ctx_data
0xcccccccccccccccc void *

Change-Id: I6f41c056d995426898228dba87e59337fc4ac356

show more ...

ded7dcfc15-Dec-2015 Tommie McAfee <tahmmee@gmail.com>

use percentage value as regression threshold

currently setting to 10% across all stats

Change-Id: I0cc3a89e75560dd697db9fad66d9c84f5dcfd124

224359f705-Jan-2016 Sundar Sridharan <sundar.sridharan@gmail.com>

MB-17215: windows crash: set larger fdb_log error msg buffer

Some error messages if too long can cause buffer overflow.
This seems to happen on windows for some reason.
Bumping up th

MB-17215: windows crash: set larger fdb_log error msg buffer

Some error messages if too long can cause buffer overflow.
This seems to happen on windows for some reason.
Bumping up the error message buffer size in fdb_log

Change-Id: I91a1e1baa330f5ca5f6c6d9cb1c55e469b77bd7d

show more ...

5e946e4204-Jan-2016 Sundar Sridharan <sundar.sridharan@gmail.com>

MB-17216: use of freed pointer for latency stat in snapshot_open causing crash

Change-Id: Idf09a3dbb70ffe6f60c20d9fb9a36ca266c3d535

9e75235323-Dec-2015 Sundar Sridharan <sundar.sridharan@gmail.com>

MB-17159: fix iterator functional test for 32 bit

storing a 64 bit offset using sizeof(pointer) is wrong for 32 bit
causes make test failures on 32 bit platforms.

Change-Id: I71

MB-17159: fix iterator functional test for 32 bit

storing a 64 bit offset using sizeof(pointer) is wrong for 32 bit
causes make test failures on 32 bit platforms.

Change-Id: I716a2a875a77e663cb93a4cccc4b04fd7cdf54ae

show more ...

2e8d77dd12-Dec-2015 Chiyoung Seo <chiyoung.seo@gmail.com>

Merge remote-tracking branch 'origin/sherlock' into master

* origin/sherlock:
Set the writer throttling delay in a less aggressive way.

Conflicts:
src/forestdb.cc

Merge remote-tracking branch 'origin/sherlock' into master

* origin/sherlock:
Set the writer throttling delay in a less aggressive way.

Conflicts:
src/forestdb.cc

Change-Id: I6785dbc70c5ff9e794255d059c8b60d25ab8ce26

show more ...


f91ff4fa11-Dec-2015 Chiyoung Seo <chiyoung.seo@gmail.com>

Set the writer throttling delay in a less aggressive way.

Setting the writer's throttling delay to 4X of the compactor's delay per doc
is too much aggressive and consequently increases t

Set the writer throttling delay in a less aggressive way.

Setting the writer's throttling delay to 4X of the compactor's delay per doc
is too much aggressive and consequently increases the writer's latency and
reduce its throughput.

From the tests, lowering the ratio improves the writer's latency and throughput
without compromising the compaction time (within 5%).
We plan to make the ratio auto-tunable as a separate improvement task.

Change-Id: I9d602b207dba39c77022761e306d30f3db01c9ca

show more ...

19ac7b3a11-Dec-2015 Chiyoung Seo <chiyoung.seo@gmail.com>

Merge remote-tracking branch 'origin/sherlock' into master

* origin/sherlock:
MB-16913 reset 'lastbid' if read fails in docio_read_through_buffer
Log more docio errors
Log

Merge remote-tracking branch 'origin/sherlock' into master

* origin/sherlock:
MB-16913 reset 'lastbid' if read fails in docio_read_through_buffer
Log more docio errors
Log more details on docio errors

Change-Id: I2e05a1e6825f632e88d802aa22a960b816f2bf44

show more ...


03716c4411-Dec-2015 Jung-Sang Ahn <jungsang.ahn@gmail.com>

MB-16913 reset 'lastbid' if read fails in docio_read_through_buffer

- Although read fails due to various reasons, the contents of
'dhandle->readbuffer' may change and contain garbage dat

MB-16913 reset 'lastbid' if read fails in docio_read_through_buffer

- Although read fails due to various reasons, the contents of
'dhandle->readbuffer' may change and contain garbage data as a result
of read failure.

- The most simple solution is to reset 'lastbid' to read the block
once again at the next _docio_read_through_buffer call.

Change-Id: Ib1f38a89240ecfedd10c8771e5c6f0d3424f3e23

show more ...

1...<<11121314151617181920>>...53