| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |\ \
| | | |
| | | |
| | | |
| | | | |
Conflicts:
src/config/config_def.c
|
| | |/
| |/|
| | |
| | | |
reference #1536, #1739.
|
| |/
|/|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
into place so a crash can't result in a corrupted base configuration
file.
Always create a base configuration file, and always include the version
number (who knows, the initial version might be useful some day).
By always creating a base configuration file, we can detect a crash
between creating the "WiredTiger" file and creating the base
configuration file, and the create the base configuration file when
we're restarted.
That's a (hopefully minor) API change: an application that initially
created the database without config_base set, or simply without any
wiredtiger_open configuration, so that no base configuration file was
written, which then ran this version of the code with config_base set
or with a wiredtiger_open configuration, would suddenly create a base
configuration file where there was none before. I find that unlikely
enough I'm not going to worry about it.
Remove the check for the base-configuration file already existing when
the database is created. A previous revision in this file (#1775) syncs
the "WiredTiger" file to disk before we create the base configuration
file, so the test be safe, but we're no longer relying on ordering
between the creation of the "WiredTiger" file and the base configuration
file, so the test no longer makes sense.
Add simple helper functions to open/close FILE handles so there's less
memory allocation and errno handling inside more complicated routines.
Reference #1775, #1776, SERVER-17571.
|
|/
|
|
| |
Still needs some test code.
|
|\
| |
| | |
Change checkpoints to do first-fit allocation
|
| |
| |
| |
| |
| | |
actual checkpoint blocks, that way if we delete significant space, the
checkpoint blocks won't prevent file truncation.
|
| | |
|
|\ \
| |/ |
|
| |\
| | |
| | | |
Fix a bug in the reconfigure API related to shared cache quotas.
|
| | |
| | |
| | |
| | | |
__cache_config_local, make it static.
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | | |
While fixing the bug simplify the code flow for reconfigure and
caches/shared caches.
refs #1712
|
| | |
| | |
| | |
| | | |
common) WT_REF flavor. Don't free instantiated keys in parent pages.
|
| |/
| |
| |
| | |
refs #1582
|
|\ \
| |/ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
The case that could return EBUSY was when checkpointing with a
specific target, while that target was open exclusively or for a bulk
load.
Refs #1404 #1589
|
| |
| |
| |
| |
| |
| |
| |
| | |
__wt_evict_file_exclusive_on/__wt_evict_file_exclusive_off pair, we
could clear WT_BTREE_NO_EVICTION on a file (not that I'm sure that
combination makes any sense, but regardless). Push the test of the
WT_BTREE_NO_EVICTION flag down into the __wt_evict_file_exclusive_on
function to make sure it doesn't happen again.
|
|/ |
|
|\
| |
| | |
Improvements to eviction for MongoDB workloads
|
| | |
|
| |
| |
| |
| | |
fix tracking of whether we are making progress.
|
| |
| |
| |
| | |
rework, take another run at it.
|
|\ \
| |/
|/| |
WT_SESSION.strerror
|
| | |
|
| | |
|
|/
|
|
|
|
|
|
|
|
| |
A feature request to allow for a high insert throughput into a
table with a size limitation.
Adds a new configuration option to WT_SESSION::create which is
lsm=(chunk_count_limit=0), default to 0 which is disabled.
Refs #1652
|
|\
| |
| | |
Struct alignment and packing for MSVC
|
| | |
|
| |
| |
| |
| |
| | |
matching the corresponding dirty-update function and getting rid of
a long-ago vestige of reconciliation's naming.
|
|\ \
| | |
| | | |
eviction statistics cleanup.
|
| | |
| | |
| | |
| | |
| | |
| | | |
cache handling into a separate function.
refs #1596
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Don't count pages evicted by a worker thread as an "application thread"
eviction; add a new statistic to distinguish between the server itself
evicting pages and the eviction worker threads evicting.
Don't increment the eviction counters unless we find a page to evict,
__evict_lru_pages() gets called a huge number of times in any workload
where eviction is happening.
Discussion:
The "is_app" flag is being used for a few purposes:
- if we should perform the "oldest transaction in the system" test,
- if the evictor thread is the eviction server,
- if the "eviction by an application thread" counter should be incremented,
- if the session's split generation can be non-zero after eviction.
Move the "oldest transaction in the system" test outside of the eviction
code, and do that test before trying to evict a page (this required
moving __wt_txn_am_oldest() from include/txn.i (an inlined function)
into txn/txn.c (a real function).
Replace the "is_app" flag with an "is_server" flag to distinguish
between the eviction server and eviction work threads (we can
distinguish between application threads and worker/server threads using
the WT_SESSION_INTERNAL flag in the WT_SESSION handle).
Reference SERVER-16997, SERVER-17020.
|
|/ /
| |
| |
| |
| |
| | |
since we've resolved the underlying problem.
This reverts commit 010059e721d1c6bd4d81214ab80402866de41345.
|
| |
| |
| |
| |
| | |
__wt_cursor_equal. We'll do a better fix to avoid the problem in the
future, when we have a few minutes.
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | | |
Conflicts:
src/cursor/cur_std.c
src/cursor/cur_table.c
src/include/extern.h
|
| |\ \
| | |/ |
|
| |\ \ |
|
| |\ \ \
| | | | |
| | | | |
| | | | |
| | | | | |
Conflicts:
src/cursor/cur_metadata.c
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
While here, add a generic cursor equals stub in curstd, rather than
redirecting via cursor function pointer lists.
Implement the Java API wrapper.
|
| |_|_|/
|/| | | |
|
|\ \ \ \
| | | | |
| | | | | |
Windows error string support.
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
os_errno.c:38:3: error: return discards 'const' qualifier from pointer
target type [-Werror]
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
possible error returns into "constant string" and other, and checking
for constant strings before anything else.
|
| | |_|/
| |/| |
| | | |
| | | |
| | | | |
which allows us to support Windows' errors where we have to allocate a
buffer to get the error string.
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
then the database is reopened with logging turned on.
refs SERVER-16692
|
|\ \ \ \
| |/ / /
|/| | | |
Fix json formatting of printlog to be valid json
|
| | | | |
|