Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Merge branch 'develop' into memory-allocation | Michael Cahill | 2015-01-27 | 1 | -0/+2 |
|\ | | | | | | | | | Conflicts: src/include/wiredtiger.in | ||||
| * | Two "real" changes: | Keith Bostic | 2015-01-23 | 1 | -0/+2 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 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. | ||||
* | | Apply the cache size adjustment when checking if the cache is full. That ↵ | Michael Cahill | 2015-01-27 | 1 | -2/+0 |
| | | | | | | | | | | | | should play nice with shared caches. refs #1590 | ||||
* | | Clean up statistics -- leave cache_bytes_max alone, create a new | Keith Bostic | 2015-01-23 | 1 | -0/+2 |
| | | | | | | | | statistic that reports the adjusted bytes. | ||||
* | | Add the cache_overhead configuration string to allow applications to configure | Keith Bostic | 2015-01-22 | 1 | -0/+1 |
|/ | | | | their cache overhead. | ||||
* | Split out statistic for forced eviction. It's not always because the | Alex Gorrod | 2015-01-13 | 1 | -0/+2 |
| | | | | page grew too large. | ||||
* | Track the largest in-memory page size we've been called on to evict. | Keith Bostic | 2015-01-10 | 1 | -0/+2 |
| | |||||
* | busy-work: sort the stat options, make the stat options units match | Keith Bostic | 2014-12-19 | 1 | -3/+3 |
| | | | | | | other stat option strings, sleep_cnt and wait_cnt should be unsigned, initialize wait_cnt to 0 and increment it in the if/test for clarity, rename page_in_sleep to page_sleep, matches other yield fields. | ||||
* | Wrong unit in description | Susan LoVerso | 2014-12-19 | 1 | -1/+1 |
| | |||||
* | Add sleep backoff if waiting for page too long. SERVER-16269 | Susan LoVerso | 2014-12-19 | 1 | -4/+5 |
| | |||||
* | Add separate statistics for pages yielding because the WT_REF isn't | Keith Bostic | 2014-12-19 | 1 | -0/+12 |
| | | | | available. | ||||
* | Merge branch 'develop' into dhandle-hash | Michael Cahill | 2014-12-16 | 1 | -4/+6 |
|\ | | | | | | | | | Conflicts: test/suite/test_sweep01.py | ||||
| * | Merge pull request #1454 from wiredtiger/overflow-relax | Michael Cahill | 2014-12-15 | 1 | -4/+6 |
| |\ | | | | | | | Separate the btree maximum key/value sizes from the underlying page size | ||||
| | * | Merge branch 'develop' into overflow-relax | Keith Bostic | 2014-12-09 | 1 | -0/+5 |
| | |\ | |||||
| | * | | Separate the btree maximum key/value sizes from the underlying page | Keith Bostic | 2014-12-09 | 1 | -4/+6 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | size, reference #1282. Deprecate the internal_item_max and leaf_item_max configuration strings, replace with internal_key_max, leaf_key_max and leaf_value_max. Remove examples/c/ex_file.c (there's no real need for a "file" URI example, and it's easy to replace the one place the documentation used it). | ||||
* | | | | Add a few more sweep stats to understand why the test sometimes fails. | Susan LoVerso | 2014-12-15 | 1 | -0/+2 |
|/ / / | | | | | | | | | | Increase test time to 120 seconds. | ||||
* | | | Add connection sweep stats and a sweep test. #1446 | Susan LoVerso | 2014-12-10 | 1 | -0/+2 |
| |/ |/| | |||||
* | | Don't compress if the record is already smaller than the minimum | Susan LoVerso | 2014-12-09 | 1 | -0/+1 |
| | | | | | | | | allocation size. #1359 | ||||
* | | Merge branch 'develop' into log-compress | Don Anderson | 2014-12-04 | 1 | -17/+17 |
|\ \ | |/ | | | | | | | Conflicts: dist/api_data.py | ||||
| * | Merge branch 'develop' into log-recycle | Michael Cahill | 2014-12-04 | 1 | -17/+17 |
| |\ | | | | | | | | | | | | | Conflicts: dist/api_data.py | ||||
| | * | Change the Python scripts to follow Python indenting standards (4 space | Keith Bostic | 2014-12-03 | 1 | -17/+17 |
| | | | | | | | | | | | | indents, spaces only). | ||||
* | | | Added log compression stats. refs #1359. | Don Anderson | 2014-12-04 | 1 | -0/+4 |
|/ / | |||||
* | | Merge branch 'develop' into log-recycle | Michael Cahill | 2014-12-02 | 1 | -0/+3 |
|\ \ | |/ | | | | | | | Conflicts: src/include/wiredtiger.in | ||||
| * | Merge branch 'develop' into split-right | Michael Cahill | 2014-12-02 | 1 | -0/+1 |
| |\ | | | | | | | | | | | | | Conflicts: src/include/wiredtiger.in | ||||
| | * | Add a statistic to count how much eviction application threads are doing. | Michael Cahill | 2014-12-02 | 1 | -0/+1 |
| | | | |||||
| * | | Add code to split a too-large insert list onto a new page as part of | Keith Bostic | 2014-11-30 | 1 | -0/+2 |
| |/ | | | | | | | | | reconciling/evicting a page, to catch append-only workloads and keep from making threads wait for the reconciliation/eviction. | ||||
* | | Merge branch 'develop' into log-recycle | Susan LoVerso | 2014-12-01 | 1 | -26/+51 |
|\ \ | |/ | |||||
| * | First pass at adding groupings for stats. #1365 | Susan LoVerso | 2014-11-26 | 1 | -27/+51 |
| | | |||||
| * | Add --all option to create an HTML file for each prefix. #1365 | Susan LoVerso | 2014-11-21 | 1 | -0/+1 |
| | | |||||
* | | Change prepare to prealloc and make it a value rather than boolean | Susan LoVerso | 2014-11-25 | 1 | -4/+3 |
| | | | | | | | | config. | ||||
* | | Just pre-allocate log files instead of recycling. #1373 | Susan LoVerso | 2014-11-24 | 1 | -6/+4 |
| | | |||||
* | | Add recycle stats and code to set a maximum number of recycle files. | Susan LoVerso | 2014-11-20 | 1 | -1/+7 |
|/ | |||||
* | Fix error wrapper usage. Change units to msecs. #1374 | Susan LoVerso | 2014-11-12 | 1 | -4/+6 |
| | |||||
* | Add checkpoint timer stats. #1374 | Susan LoVerso | 2014-11-12 | 1 | -0/+10 |
| | |||||
* | Rename the log "user bytes" statistic to "payload bytes". The former has ↵ | Michael Cahill | 2014-11-10 | 1 | -1/+1 |
| | | | | confused a few people, I hope the new wording is clearer. | ||||
* | Get rid of some abbreviations in the async statistics messages. | Keith Bostic | 2014-11-03 | 1 | -9/+10 |
| | |||||
* | Now that the tag is part of the description, the right field to sort | Keith Bostic | 2014-11-03 | 1 | -1/+1 |
| | | | | on is the description, not the name. | ||||
* | Make capitalization consistent. | Keith Bostic | 2014-11-03 | 1 | -3/+3 |
| | |||||
* | whitespace | Keith Bostic | 2014-11-03 | 1 | -3/+3 |
| | |||||
* | Replace some of the tags with more verbose versions: | Keith Bostic | 2014-11-03 | 1 | -4/+4 |
| | | | | | | | | | | | conn -> connection, dhandle -> data-handle txn -> transaction Don't use whitespace in the tag name so post-processing is a little simpler: block manager -> block-manager | ||||
* | Use the "btree" tag consistently (before we had both "Btree" and "btree"). | Keith Bostic | 2014-11-03 | 1 | -28/+25 |
| | |||||
* | Create a set of subclasses to ensure we can't forget prefixes or create | Keith Bostic | 2014-11-03 | 1 | -289/+289 |
| | | | | | | | | | | inconsistent ones. This change exactly matches the previous output, I want to make sure it's clear where I've changed the output vs. where I'm changing how we create that output. Reference #1313. | ||||
* | unused statistics field: rec_skipped_update | Michael Cahill | 2014-11-03 | 1 | -2/+0 |
| | |||||
* | In the case of a workload where we're forcibly evicting a large page, | Keith Bostic | 2014-11-01 | 1 | -3/+0 |
| | | | | | | | but most of the page is discarded during reconciliation and the page doesn't split into multiple chunks, we were quitting, leaving the page in place. Instead, instantiate the page and swap it into place to replace the previous version. Reference #1317. | ||||
* | Fix whitespace in dist/stat_data.py | Michael Cahill | 2014-10-28 | 1 | -364/+364 |
| | |||||
* | Add prefixes to all stat descriptions. #1313 | Susan LoVerso | 2014-10-27 | 1 | -122/+147 |
| | |||||
* | Add statistics to track open sessions and the range of transaction IDs that ↵ | Michael Cahill | 2014-10-24 | 1 | -0/+4 |
| | | | | are pinned. | ||||
* | If we wrap a log file before the previous one closes, wait instead of | Susan LoVerso | 2014-10-21 | 1 | -0/+2 |
| | | | | asserting failure. #1302 | ||||
* | Don't list the potential object types, it doesn't add information. | Keith Bostic | 2014-10-06 | 1 | -2/+1 |
| | |||||
* | Add stat to count number of times we reach queue_ref maximum. #1230 | Susan LoVerso | 2014-09-18 | 1 | -0/+1 |
| |