| Commit message (Expand) | Author | Age | Files | Lines |
* | Update copyright notice to 2013. | Keith Bostic | 2013-01-01 | 31 | -31/+31 |
* | Rename __evict_list_clr_all to __evict_list_clr_range, it no longer clears | Keith Bostic | 2012-12-26 | 1 | -5/+5 |
* | If a page splits, it potentially creates a merge-split internal page and we | Keith Bostic | 2012-12-23 | 1 | -0/+10 |
* | Now that __wt_cache_page_inmem_incr() does atomic updates, there's no | Keith Bostic | 2012-12-21 | 2 | -9/+9 |
* | Don't check the page's write-generation number: we're going to check the | Keith Bostic | 2012-12-19 | 1 | -6/+5 |
* | lint complained, __evict_readonly return ignored -- changed to void. | Keith Bostic | 2012-12-19 | 1 | -3/+1 |
* | Merge pull request #413 from wiredtiger/concurrent-checkpoints | Michael Cahill | 2012-12-18 | 3 | -74/+250 |
|\ |
|
| * | Address Keith's review feedback on concurrent checkpoints. | Michael Cahill | 2012-12-19 | 3 | -125/+188 |
| * | Merge branch 'develop' into concurrent-checkpoints | Michael Cahill | 2012-12-18 | 3 | -18/+15 |
| |\ |
|
| * | | Wake the eviction server when dirty eviction is enabled. | Michael Cahill | 2012-12-18 | 1 | -1/+8 |
| * | | Make sure there is no LRU eviction underneath an internal node during checkpo... | Michael Cahill | 2012-12-18 | 1 | -5/+9 |
| * | | Skip dirty leaf pages during the locked phase of a checkpoint. They have alr... | Michael Cahill | 2012-12-18 | 1 | -3/+20 |
| * | | Add a "wait" mode to walking the cache using hazard pointers instead of the "... | Michael Cahill | 2012-12-18 | 2 | -10/+23 |
| * | | Add a "pre-checkpoint" pass that writes dirty leaf pages without interrupting... | Michael Cahill | 2012-12-18 | 2 | -4/+11 |
| * | | Fixes from Alex's review: revert a test change made for debugging, and switch... | Michael Cahill | 2012-12-18 | 1 | -3/+5 |
| * | | Increase the scope of read-only eviction until after the checkpoint is written. | Michael Cahill | 2012-12-17 | 1 | -6/+14 |
| * | | Always re-enable eviction of dirty pages after a checkpoint (not just on error). | Michael Cahill | 2012-12-14 | 1 | -5/+5 |
| * | | Deal with deleted pages in new checkpoint walk. | Michael Cahill | 2012-12-14 | 1 | -4/+4 |
| * | | When evicting in a file that is being checkpointed, only evict clean pages. | Michael Cahill | 2012-12-14 | 1 | -5/+14 |
| * | | Try to clear the LRU walk if waiting in the tree during a checkpoint. | Michael Cahill | 2012-12-14 | 2 | -27/+35 |
| * | | Perform checkpoints in the calling thread, don't block eviction. | Michael Cahill | 2012-12-13 | 3 | -20/+58 |
* | | | Merge pull request #414 from wiredtiger/concurrent-updates | Michael Cahill | 2012-12-18 | 2 | -14/+22 |
|\ \ \ |
|
| * | | | Implement review feedback. | Michael Cahill | 2012-12-19 | 1 | -6/+5 |
| * | | | Merge branch 'develop' into concurrent-updates | Keith Bostic | 2012-12-18 | 4 | -35/+21 |
| |\ \ \
| | | |/
| | |/| |
|
| * | | | Allow non-conflicting updates to complete. | Michael Cahill | 2012-12-17 | 2 | -12/+21 |
* | | | | Wrap __wt_panic uses in a macro. Update LSM panic code to be better | Alex Gorrod | 2012-12-19 | 1 | -4/+2 |
| |/ /
|/| | |
|
* | | | Remove HAVE_VERBOSE and most of the verbose convenience macros, verbose | Keith Bostic | 2012-12-17 | 4 | -35/+21 |
|/ / |
|
* | | Make the tree build without HAVE_VERBOSE. | Keith Bostic | 2012-12-14 | 4 | -22/+33 |
|/ |
|
* | Document why we're ignoring the return from WT_BTREE_CMP in one case. | Keith Bostic | 2012-12-12 | 1 | -0/+6 |
* | Don't ignore the possible error return from WT_BTREE_CMP. | Keith Bostic | 2012-12-12 | 1 | -1/+1 |
* | Don't ignore the return value of __rec_write_wrapup_err. | Keith Bostic | 2012-12-12 | 1 | -1/+1 |
* | Now some errors can override other errors, we shouldn't ignore errors | Keith Bostic | 2012-12-12 | 2 | -2/+2 |
* | We have a panic function: make a failure to clear a hazard pointer | Keith Bostic | 2012-12-12 | 11 | -38/+40 |
* | Panic if the eviction server unexpectedly exits. | Keith Bostic | 2012-12-10 | 1 | -1/+3 |
* | Rename WT_SERVER_RUN to be WT_CONN_SERVER_RUN to match all the other | Keith Bostic | 2012-12-10 | 1 | -2/+2 |
* | Merge pull request #408 from wiredtiger/row-key-serial | Michael Cahill | 2012-12-09 | 1 | -42/+16 |
|\ |
|
| * | Replace the row-key-serial function with CAS instruction. | Keith Bostic | 2012-12-07 | 1 | -42/+16 |
* | | Merge pull request #405 from wiredtiger/mtx-cleanup | agorrod | 2012-12-09 | 6 | -17/+15 |
|\ \ |
|
| * | | Change __wt_readlock, __wt_try_writelock, __wt_writelock, __wt_rwunlock | Keith Bostic | 2012-12-07 | 2 | -5/+5 |
| * | | Change __wt_cond_wait and __wt_cond_signal to return error values, | Keith Bostic | 2012-12-07 | 4 | -12/+10 |
| |/ |
|
* | | We document the key-gap can be set to 0 (which makes logical sense, a gap of | Keith Bostic | 2012-12-08 | 1 | -2/+4 |
|/ |
|
* | Merge branch 'develop' into doc_mods1 | Susan LoVerso | 2012-12-06 | 11 | -72/+75 |
|\ |
|
| * | Split cache_bytes_changed into cursor_{insert,remove,update}_bytes and | Keith Bostic | 2012-12-06 | 1 | -3/+3 |
| * | Push further down the "data-source" path: rename WT_BSTAT_XXX to | Keith Bostic | 2012-12-06 | 10 | -57/+57 |
| * | Rename btree_entries_bulk_loaded to be cursor_insert_bulk, we don't | Keith Bostic | 2012-12-06 | 1 | -1/+1 |
| * | Use WT_CSTAT_INCR to reference connection statistics. | Keith Bostic | 2012-12-06 | 1 | -2/+1 |
| * | Review the statistics: rename the statistics names for consistency and | Keith Bostic | 2012-12-05 | 10 | -56/+60 |
* | | Change hazard reference to hazard pointer, plus a few minor doc edits. | Susan LoVerso | 2012-12-06 | 7 | -15/+15 |
|/ |
|
* | Fix a potential bug in dirty page tracking. Loosen constraints on | Alex Gorrod | 2012-12-03 | 2 | -3/+4 |
* | Fix logic error in eviction loop. | Michael Cahill | 2012-11-30 | 1 | -3/+5 |