summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Cut release 1.6.11.6.1Alex Gorrod2013-05-316-10/+45
|
* Link custom data sources documentation into the docs.Alex Gorrod2013-05-311-0/+1
|
* Have test/format sleep for the calculated period between hot backups.Michael Cahill2013-05-311-6/+5
|
* Remove unused variables.Michael Cahill2013-05-311-2/+0
|
* Fix a leak in the free-on-close list code used for extension-specific config.Michael Cahill2013-05-311-22/+15
|
* Fix a core dump, the realloc has to change to match the backup cursor'sKeith Bostic2013-05-301-2/+4
| | | | new structure sizing.
* Merge pull request #556 from wiredtiger/hot-backupAlex Gorrod2013-05-3031-514/+937
|\ | | | | | | Fix several bugs in hot backup, including race conditions between backup and table drop (and other schema level operations). @closes #557
| * fix up a couple of compiler warnings.Keith Bostic2013-05-301-3/+3
| |
| * Don't use u_int in the API, it's a non-standard type.Keith Bostic2013-05-301-2/+2
| |
| * Don't use u_int in the API, it's a non-standard type. Closes #561.Keith Bostic2013-05-304-7/+15
| |
| * We have 3 backup tests now, name them consistently.Keith Bostic2013-05-301-0/+0
| |
| * Change hot-backup to acquire a handle on each underlying file object to ensureKeith Bostic2013-05-304-42/+61
| | | | | | | | | | schema level operations (for example, drop), will be blocked until the backup cursor closes.
| * whitespaceKeith Bostic2013-05-301-1/+1
| |
| * Merge branch 'develop' into hot-backupKeith Bostic2013-05-2912-120/+166
| |\ | |/ |/|
* | remove WT_CONNECTION_IMPL.{msgfile,msgcall}, they're unused.Keith Bostic2013-05-291-3/+0
| |
* | Merge pull request #538 from wiredtiger/warm-cache-compressAlex Gorrod2013-05-299-81/+127
|\ \ | | | | | | Preload internal pages into file system cache with compression.
| * \ Merge pull request #559 from wiredtiger/warm-cache-generalAlex Gorrod2013-05-2811-87/+129
| |\ \ | | | | | | | | Create a more general-purpose block-manager preload call
| | * | Create a more general-purpose block-manager preload call and pre-loadKeith Bostic2013-05-2811-87/+129
| |/ / | | | | | | | | | | | | individual pages instead of using the page locations to guess at the right file chunks.
| * | Merge branch 'develop' into warm-cache-compressKeith Bostic2013-05-2885-647/+1235
| |\ \ | | | | | | | | | | | | | | | | Conflicts: src/include/flags.h
| * | | Preload internal pages into file system cache with compression.Alex Gorrod2013-05-144-13/+17
| | | | | | | | | | | | | | | | Previously the functionality only worked without compression enabled.
* | | | Another transaction ID fix: ignore the current ID if updating a snapshot ↵Michael Cahill2013-05-291-5/+9
| | | | | | | | | | | | | | | | during auto-commit.
* | | | Re-check that calculating an ordinary snapshot hasn't raced with updating ↵Michael Cahill2013-05-291-1/+1
| | | | | | | | | | | | | | | | the oldest ID.
* | | | Build the connection config stack explicitly, fix a bug where the ↵Michael Cahill2013-05-291-30/+29
| |/ / |/| | | | | | | | environment config was appearing at the beginning.
* | | Add upgrade documentation for new shared_cache configuration option.Alex Gorrod2013-05-281-0/+7
| | |
| | * Fix a bug where LSM would get the same lock twice during a drop.Alex Gorrod2013-05-291-10/+22
| | |
| | * Add upgrade documentation for new shared_cache configuration option.Alex Gorrod2013-05-281-0/+7
| | |
| | * Merge branch 'develop' into hot-backupAlex Gorrod2013-05-2816-227/+252
| | |\ | |_|/ |/| |
* | | Add an explicit enable setting to shared_cache configuration.Alex Gorrod2013-05-287-16/+18
| | | | | | | | | | | | | | | | | | | | | | | | It used to rely on the shared_cache configuration group being present in the user configuration string. This change will require applications using the shared cache functionality to add "enable=true" to the shared_cache configuration group.
* | | Merge pull request #552 from wiredtiger/force-evict-indexAlex Gorrod2013-05-289-211/+234
|\ \ \ | | | | | | | | Fixed forced eviction with indexes.
| * | | Remove a bogus assertion from transaction ID management.Michael Cahill2013-05-281-3/+1
| | | |
| * | | Avoid a race when updating the shared table of transaction IDs and ↵Michael Cahill2013-05-282-81/+91
| | | | | | | | | | | | | | | | calculating the oldest active transaction ID.
| * | | Simplify scans through the global table of transaction IDs.Michael Cahill2013-05-277-200/+123
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Stop caching multiple copies of the oldest transaction ID required to stay in memory. In the process, fix a race that could invalidate the calculation of the oldest ID. refs #552
| * | | Make sure the oldest transaction ID we need always moves forward.Michael Cahill2013-05-241-3/+5
| | | |
| * | | Merge branch 'develop' into force-evict-indexAlex Gorrod2013-05-2428-98/+233
| |\ \ \
| * | | | Update index performance test to test combinations of file and LSM.Alex Gorrod2013-05-241-29/+36
| | | | |
| * | | | Update new test case to run in a reasonable time.Alex Gorrod2013-05-241-37/+13
| | | | |
| * | | | Yet another rearrangement of the forced eviction code. Now check the first ↵Michael Cahill2013-05-245-16/+28
| | | | | | | | | | | | | | | | | | | | time we read a page in a transaction. That way, it shouldn't contain updates from the current transaction that prevent eviction.
| * | | | Add a test case for a performance regression.Alex Gorrod2013-05-241-0/+95
| | | | |
| | | | * Merge pull request #558 from wiredtiger/hot-backup-schemaAlex Gorrod2013-05-2714-300/+413
| | | | |\ | | | | | | | | | | | | Update backup to use __wt_schema_worker. Stop LSM drops during backups.
| | | | | * Update __wt_schema_worker so that it can operate on names.Alex Gorrod2013-05-287-48/+51
| | | | | | | | | | | | | | | | | | | | | | | | It can now traverse either names, or files.
| | | | | * Minor code shuffling/simplification, unused variable cleanups, comment fixing.Keith Bostic2013-05-271-18/+8
| | | | | |
| | | | | * We have to write all of the names/value pairs out to the backup file,Keith Bostic2013-05-273-22/+36
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | not just the file: objects, don't discard the trailng ".wt" from the file object name, whatever is in the metadata file is what we use.
| | | | | * Re-work the code to build the target list, uncomment object checking,Keith Bostic2013-05-271-14/+19
| | | | | | | | | | | | | | | | | | | | | | | | this now works.
| | | | | * Delete test_illegal_objects, now that we're using schema_worker,Keith Bostic2013-05-271-8/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | there's no special path for illegal objects and no reason to do any special testing.
| | | | | * avoid spelling complaintKeith Bostic2013-05-271-1/+1
| | | | | |
| | | | | * Merge branch 'develop' into hot-backup-schemaKeith Bostic2013-05-274-109/+159
| | | | | |\ | |_|_|_|_|/ |/| | | | |
* | | | | | Merge pull request #555 from wiredtiger/turtle-raceMichael Cahill2013-05-264-109/+159
|\ \ \ \ \ \ | | | | | | | | | | | | | | Make the existence of a turtle file imply a known-good metadata file.
| * \ \ \ \ \ Merge branch 'develop' into turtle-raceKeith Bostic2013-05-252-3/+3
| |\ \ \ \ \ \ | |/ / / / / / |/| | | | | |
| * | | | | | We were creating the turtle file, and if it didn't previously exist we'd createKeith Bostic2013-05-254-109/+159
| | |/ / / / | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | the metadata file. If we crashed between those two steps, we'd fail the next open because the turtle file existed but the metadata file didn't. Restructure so we don't create the turtle file until we finish creating the metadata file (and loading any hot backup), that way if the turtle file exists, everything else is known to be good. Ref #554.
| | | | | * path variable no longer used, don't return without freeing the scratch buffer.Keith Bostic2013-05-271-9/+4
| | | | | |