History log of /6.0.3/forestdb/ (Results 226 - 250 of 1302)
Revision (<<< Hide revision tags) (Show revision tags >>>)Date Author Comments
(<<< Hide modified files)
(Show modified files >>>)
d5dd78a303-Feb-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

Fix to the minor memory leak caused by cancelling the compaction

Change-Id: I9fb9aa2a5a47e1e311e3f9f321e3bad4e8e09586

02fb65e403-Feb-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

MB-17810 Don't include partially committed items in in-memory snapshot

If there are any partially committed items in WAL, then they shouldn't
be copied to the in-memory snapshot. This ch

MB-17810 Don't include partially committed items in in-memory snapshot

If there are any partially committed items in WAL, then they shouldn't
be copied to the in-memory snapshot. This change makes sure that if a WAL
item belongs to any of active transactions, then it won't be included in
the in-memory snapshot requested.

Change-Id: Ibf4fbdac2675039de0aa613bbb50191f1c4a13df

show more ...

8e1e87f303-Feb-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Mark transaction commit marker as stale

- Since transaction commit marker is used only for crash recovery
(or WAL reconstruction), we can mark it as stale immediately.

- Stale r

Mark transaction commit marker as stale

- Since transaction commit marker is used only for crash recovery
(or WAL reconstruction), we can mark it as stale immediately.

- Stale regions by those markers will be inserted into stale-tree
at the next WAL flushing commit, thus will be reclaimed when the
corresponding commit header becomes unreachable. After that, those
commit markers becomes unnecessary for both crash recovery and WAL
restore.

Change-Id: Ic74f7d9db1c6bcd6ce287fba280c74ef1b070ba7

show more ...

029c71f202-Feb-2016 Chiyoung Seo <chiyoung@couchbase.com>

Merge "Do not return error if there is no DB header."


589ab73602-Feb-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Do not return error if there is no DB header.

- If a file is created & closed without any mutation, we cannot open
the file anymore.

Change-Id: I24615fac822025298c841901f31cb687

Do not return error if there is no DB header.

- If a file is created & closed without any mutation, we cannot open
the file anymore.

Change-Id: I24615fac822025298c841901f31cb68790b65f97

show more ...

d06007c428-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

MB-17484 Check all file and KV handles upon fdb_kvs_remove and fdb_rollback

When fdb_kvs_remove or fdb_rollback_all API is invoked, all the file handles
and their KV handles created shou

MB-17484 Check all file and KV handles upon fdb_kvs_remove and fdb_rollback

When fdb_kvs_remove or fdb_rollback_all API is invoked, all the file handles
and their KV handles created should be checked to return an appropriate
error to the caller if any conflict handles are still opened.

Change-Id: I06eff2f2fa7db6773e29f58e9aab3edfb8104428

show more ...

bb71615f21-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

Support configurable daemon compaction interval per file at runtime

With this change, the client can change the daemon compaction interval
for a given file at runtime once it is opened.

Support configurable daemon compaction interval per file at runtime

With this change, the client can change the daemon compaction interval
for a given file at runtime once it is opened.
Note that when a given file is opened for the first time, then its
daemon compaction interval is set by the global config param.

Change-Id: I5e5c87eb1df014484e67d41bd05fb6796a814913

show more ...

431c540d21-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

Schedule a daemon compaction task per file based on its last compaction time

As each file is compacted at a different time by a daemon compaction thread,
The file's last compaction time

Schedule a daemon compaction task per file based on its last compaction time

As each file is compacted at a different time by a daemon compaction thread,
The file's last compaction time should be considered to determine if the next
compaction should be executed or not.

Change-Id: I7e08410ab6b5346b7d59cf7bbfb3350b7dd18d51

show more ...

f6642c4e21-Jan-2016 Chiyoung Seo <chiyoung.seo@gmail.com>

MB-17384 Make block reusing parameters configurable at runtime

With this change, block_reusing_threshold and num_keeping_headers
parameters can be tuned at runtime.

Change-Id: I

MB-17384 Make block reusing parameters configurable at runtime

With this change, block_reusing_threshold and num_keeping_headers
parameters can be tuned at runtime.

Change-Id: I5864eae6882e7a90d670f9fa7ad331d3ae98089e

Conflicts:
src/forestdb.cc

show more ...

da9194a101-Feb-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix some potential bugs on reading old DB file

- Added several version handling routines.

Change-Id: I5e5a153a2f64b6ff9f5932e9368b3753401b6953

c50e978b29-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Read DB header at the end of DB file if superblock has not been written

- If DB file is closed without calling fdb_close() before the first
update of superblock, ForestDB regards it as a

Read DB header at the end of DB file if superblock has not been written

- If DB file is closed without calling fdb_close() before the first
update of superblock, ForestDB regards it as a corrupted DB file so
that all data are lost.

- To address this issue, if superblock does not contain any DB header
info, read DB header at the end of the DB file.

Change-Id: I8348bb1867ad34f4639b72e5653a737f220b15bb

show more ...

82ac339427-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Add a global index for opened FDB file handles

- Since a filemgr handle can have multiple FDB file handles, we have
to check all opened FDB file handles belonging to the same file when

Add a global index for opened FDB file handles

- Since a filemgr handle can have multiple FDB file handles, we have
to check all opened FDB file handles belonging to the same file when
getting the smallest active header revision number.

Change-Id: I0aa71b0fcff244d623199ea904ff08707df2bc2c

show more ...

98e47afe26-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix a bug skipping removal of seq numbers in single KVS mode

- In single KVS mode, kv_info_offset is always BLK_NOT_FOUND so that
we have to handle that case separately. If we don't, sta

Fix a bug skipping removal of seq numbers in single KVS mode

- In single KVS mode, kv_info_offset is always BLK_NOT_FOUND so that
we have to handle that case separately. If we don't, stale seq number
removal is not executed in single KVS mode so that live data size
is getting larger over time.

Change-Id: Ia5378e0826f9d1c83982ff03e4d17ef817ebbc15

show more ...

60b03adb25-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Remember next doc block ID during WAL restore

- If blocks are reused in a circular manner, some doc blocks can be
missed when there are more than one writer thread.

Change-Id: I

Remember next doc block ID during WAL restore

- If blocks are reused in a circular manner, some doc blocks can be
missed when there are more than one writer thread.

Change-Id: I53c2b66a5c1bd08e8c5aec62cef6e2e65c84406f

show more ...

8792e64f22-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix a bug that takes dirty seq number during snapshot open

- Even when block reuse is enabled, we should ignore dirty seq number
when there are newly allocated blocks since the last comm

Fix a bug that takes dirty seq number during snapshot open

- Even when block reuse is enabled, we should ignore dirty seq number
when there are newly allocated blocks since the last committed header.

Change-Id: Iab0fe57fa701372142d1fdf7bf3abda4c81c48c1

show more ...

81b8f64a20-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix a bug that allocates a new block beyond the bitmap size

Change-Id: I85555dbfc33d1a1b998aa1a837a14d2d7c45e9a1

2a1f5d9015-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Return reusable blocks to stale tree before compaction

- Current implementation reports a false write error if following
conditions are satisfied:

1) Some blocks are reclaimed

Return reusable blocks to stale tree before compaction

- Current implementation reports a false write error if following
conditions are satisfied:

1) Some blocks are reclaimed for reuse.
2) Compaction is invoked.
2-1) At the beginning of compaction, filemgr_commit() is called so
that a new DB header is appended at the end of the file.
2-2) The file status is changed to COMPACT_OLD; now this
file does not reuse stale blocks so that new blocks are
appended at the end of the file.
3) The compaction fails.
3-1) Now the file status is rolled back to NORMAL, and free blocks
are allocated from the reusable block pool again.
4) However, the BID of the last committed header (which was created
at 2-1)) is greater than all reusable blocks, thus ForestDB
regards those newly allocated blocks as immutable, and reports
a write error.

- To address this issue,

1) All remaining reusable blocks are returned into stale tree (of the
old file) before compaction starts.
2) Although compaction fails so that the file status becomes NORMAL
again, new block is allocated at the end of the file as reusable
block pool became empty. All those new blocks are writable since
their BID is greater than the last commit BID.

- Note that returned stale blocks will be reclaimed again sometime.

Change-Id: Ied2358774c2d501111e72a02caa666d63daf1902

show more ...

1a6d57c613-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix a crash during reopening DB file

- BMP index should be reconstructed on not allocated reusable blocks
only.

- WAL reconstruction should stop when it meets the last committed

Fix a crash during reopening DB file

- BMP index should be reconstructed on not allocated reusable blocks
only.

- WAL reconstruction should stop when it meets the last committed
header.

Change-Id: Ic34ba681dcf9737691129e7723ea25fe915de92e

show more ...

3af54d0313-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Prevent switching reserved block inside sb_rsv_append_doc()

- Switching releases sb->rsv_bmp structure so that sb_rsv_append_doc
may cause crash.

Change-Id: I97b53700cdec2cda643

Prevent switching reserved block inside sb_rsv_append_doc()

- Switching releases sb->rsv_bmp structure so that sb_rsv_append_doc
may cause crash.

Change-Id: I97b53700cdec2cda643257de5848d7838013ed68

show more ...

626308e913-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Remove dirty WAL flushing in auto commit mode

- Since _fdb_commit() is internally called soon, we don't need to
flush WAL in a dirty manner.

- This patch will improve the overal

Remove dirty WAL flushing in auto commit mode

- Since _fdb_commit() is internally called soon, we don't need to
flush WAL in a dirty manner.

- This patch will improve the overall performance of ForestDB in
auto commit mode.

Change-Id: I7dd17879047acda1d59753944b340d5f6f8b66b1

show more ...

c32ab13e12-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Do not append bitmap doc if no reusable block is reserved.

- If there is no reusable block among stale blocks, 'sb->rsv_bmp' sets
to NULL so that we should check if this field is NULL or

Do not append bitmap doc if no reusable block is reserved.

- If there is no reusable block among stale blocks, 'sb->rsv_bmp' sets
to NULL so that we should check if this field is NULL or not before
accessing the bitmap.

Change-Id: I397619bc5525a8850983f9edf4c4b62bdb9d7ab2

show more ...

8c43db9712-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Get rid of time limit for stale block reclaiming

- If stale regions are randomly scattered over DB file, partial
scanning due to time limit cannot make enough amount of reusable
bloc

Get rid of time limit for stale block reclaiming

- If stale regions are randomly scattered over DB file, partial
scanning due to time limit cannot make enough amount of reusable
blocks.

Change-Id: I121f162980173740579b2917a5dbd08b6f6338de

show more ...

5dd2198e11-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix a bug on storing bmp revnum in document block meta

- Since superblock's bmp revnum is dynamically changed over time,
doc handle should remember the last bmp revnum at the time that t

Fix a bug on storing bmp revnum in document block meta

- Since superblock's bmp revnum is dynamically changed over time,
doc handle should remember the last bmp revnum at the time that the
last block was allocated.

Change-Id: Ib31a46a75c3443ae8dcc50e08b80a8f1142ba50a

show more ...

736747c011-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Fix a bug on adding stale region when a sub-block is enlarged

- This bug causes the stale block leakage of index nodes.

Change-Id: Ia30e6db1bc3435c7806f0a61abbff36951709919

27376f3b05-Jan-2016 Jung-Sang Ahn <jungsang.ahn@gmail.com>

Pre-reclaim stale blocks for next round block reuse

- Reclaiming stale blocks only when there is no more free block makes
the file size increase more and more, as there is some time gap

Pre-reclaim stale blocks for next round block reuse

- Reclaiming stale blocks only when there is no more free block makes
the file size increase more and more, as there is some time gap between
the exhaustion of free blocks and reclaiming new blocks.

- To address this issue, we need to pre-reclaim stale blocks for the
next round block reuse.

Change-Id: I4b8092c81038a51ed786a90e97560e5d228705b2

show more ...

12345678910>>...53