158 check-ins using file TODO version fa3d981ca6
|
2023-05-24
| ||
| 18:39 | guarded a create-directory with directory-exists? check-in: 60290c5f00 user: mmgraham tags: v1.80 | |
| 04:16 | Added exception handlers on transactions on sync and in one additional with-input in simple locks. check-in: 9062c1c10d user: matt tags: v1.80 | |
|
2023-05-23
| ||
| 20:48 | Simplify the locking scenario for sync check-in: cdf8c77efe user: matt tags: v1.80 | |
| 19:19 | Fixed attach sync check-in: a8fa1eb8a2 user: matt tags: v1.80 | |
|
2023-05-22
| ||
| 21:38 | Couple untested fixes check-in: 4a2131ba1d user: matt tags: v1.80 | |
| 20:41 | One failure to get lock to open db go ahead and try, reset counters of api calls using count of threads, some threads might die and are unable to unregister themselves. check-in: e799a787ef user: matt tags: v1.80 | |
| 17:50 | Cherry pick ab049 check-in: 64f486e05b user: matt tags: v1.80 | |
| 17:21 | Missed change check-in: 63f66b4145 user: matt tags: v1.80 | |
| 17:00 | Cherry pick 1443 and 41255, caching check-in: 816d0a281b user: matt tags: v1.80-servload2 | |
| 16:52 | Added server parameter debug-parameter (set to -:p to profile for example). check-in: d15b736af8 user: matt tags: v1.80 | |
| 12:09 | Changed delay to linear function 0.25 * number of threads over 3. check-in: f42bab9286 user: matt tags: v1.80 | |
| 11:47 | Added exception handlers back - within threads they have to be dealth with. Added handling of busy exception and it seems to be working better (buy not enough testing yet to be sure). check-in: 6c641b6f76 user: matt tags: v1.80 | |
| 10:32 | Remove some exception handlers that were not fully specified check-in: 2f04a0e3c6 user: matt tags: v1.80 | |
| 08:37 | Few tweaks to be a bit more resiliant on database blockages. Root cause of blockages are not known yet. check-in: b55a88229c user: matt tags: v1.80 | |
| 06:13 | Turned back on the server exit on transport fail. Few other changes. check-in: fc272da6d4 user: matt tags: v1.80 | |
| 05:04 | Removed attempt to exit server on handler failure. It did not seem to be working. check-in: 47cfd09de7 user: matt tags: v1.80 | |
|
2023-05-21
| ||
| 20:44 | Check for sync in progress before even launching megatest -db2db check-in: 4579b102a3 user: matt tags: v1.80 | |
| 07:44 | Added timed ping but not used yet for throttling. check-in: 99d5a8b353 user: matt tags: v1.80 | |
| 07:20 | Use busy more aggressively and turn off loaded in tcp-transport. check-in: 78c69ec26e user: matt tags: v1.80 | |
|
2023-05-20
| ||
| 22:03 | If busy respond to ping as per normal. check-in: 1b2c38a5ef user: matt tags: v1.80 | |
| 21:49 | Turn busy throttle back on but with fixed 15 second delay. check-in: 218ab3d565 user: matt tags: v1.80 | |
|
2023-05-14
| ||
| 20:36 | Formalize /tmp directory creation for consistent usage and ultimately to accomodate second sync to dashboard on readonly check-in: 9a780869ba user: matt tags: v1.80 | |
|
2023-05-13
| ||
| 22:37 | Remove traces of inmem check-in: 5a507eea4f user: matt tags: v1.80 | |
| 20:11 | key is todisk, not todest. Although todest would be more sensible but the attach sync prevents being sensible. check-in: de7cca3a62 user: matt tags: v1.80 | |
| 00:27 | Turned off sync when direct-to-db mode 'none is used check-in: 49befe770d user: matt tags: v1.80 | |
| 00:00 | Fixed server lock. Noticed sync running from dashboard only - seems wrong. check-in: ac41427eb2 user: matt tags: v1.80 | |
|
2023-05-11
| ||
| 12:12 | Gate running the megatest syncer by existance of lock file. check-in: 12c06b6095 user: matt tags: v1.80 | |
| 09:14 | Merged fork check-in: b21843154b user: mrwellan tags: v1.80 | |
| 09:12 | Added useful target for commits in Makefile check-in: 72d16d9632 user: mrwellan tags: v1.80 | |
| 09:08 | reworked the sync locking, allow parallel servers (needs work to lock in to single machine) check-in: 7fb44b797e user: matt tags: v1.80 | |
| 05:43 | Use a key for the db lock-down that is unique to the db in .mtdb, this should elminate duplicate, overlapping servers. check-in: 138a40d18e user: matt tags: v1.80 | |
|
2023-05-10
| ||
| 22:34 | Cherrypicked several changes from v1.70 check-in: 3ba90485f1 user: mmgraham tags: v1.80 | |
| 20:33 | Patched in the -db2db code and it appears to work fine. check-in: 0131a588a0 user: matt tags: v1.80 | |
|
2023-05-08
| ||
| 05:23 | Lock defaults of tmp for cache and none for dashboard check-in: 08314706e4 user: matt tags: v1.80 | |
|
2023-05-06
| ||
| 18:11 | Added serialize-env back in Makefile. Stabilize ids on init of keys and MEGATEST_VERSION on db creation (from ab0494b4b681fb8fe2729874147a548bcb009fa9) check-in: d06c952405 user: matt tags: v1.80 | |
| 16:39 | Turn off find-and-mark-incomplete fully check-in: 3f75938dff user: matt tags: v1.80 | |
|
2023-04-17
| ||
| 23:12 | Caching extended to couple more calls. Impact seems dramatic. However note that likely not all scenarios for clearing the cache are covered. check-in: 4126673c03 user: matt tags: v1.80-servload | |
| 17:28 | cache get-test-info-by-id check-in: 1443998a16 user: matt tags: v1.80-servload | |
| 16:51 | possible fixes to test Leaf check-in: 4ef17eb32c user: matt tags: v1.80-possible-fixes | |
|
2023-04-16
| ||
| 16:43 | possible fix for bind issue check-in: 66f2b72697 user: matt tags: v1.80 | |
|
2023-04-15
| ||
| 09:14 | merged fork check-in: 532a2581c9 user: matt tags: v1.80 | |
|
2023-04-14
| ||
| 12:56 | committed merge check-in: 4131c9a1b5 user: mmgraham tags: v1.80-servload-side | |
| 06:17 | Missed a location using launch:is-test-alive, converted it to commonmod: check-in: c7687dbe89 user: matt tags: v1.80 | |
|
2023-04-13
| ||
| 15:54 | added assert for good db to dbmod:with-db check-in: b6788d6803 user: mmgraham tags: v1.80 | |
| 15:37 | Removed assert that is in wrong place check-in: f026e9979d user: mrwellan tags: v1.80 | |
| 13:12 | removed debug print check-in: 959e918725 user: pjhatwal tags: v1.80 | |
| 10:51 | Fix for servinffile check-in: f540745977 user: pjhatwal tags: v1.80 | |
| 07:01 | Turn back on update on db touched. check-in: 5cad598e0e user: matt tags: v1.80 | |
|
2023-04-12
| ||
| 20:44 | Reverted dashboard to v1.65 code and made only necessary changes Leaf check-in: eb37de4dab user: matt tags: v1.80-v1.65-dashboard | |
| 18:17 | guarded against non-initialized tt-servinf-file check-in: f971fe9c6c user: mmgraham tags: v1.80 | |
| 12:50 | moved mtargs before debugprint check-in: a8323ba69f user: pjhatwal tags: v1.80 | |
| 10:19 | merged fork check-in: 5d3a86162e user: mmgraham tags: v1.80 | |
| 08:47 | Limit backoff delay, check that param is a condition check-in: ff60788828 user: matt tags: v1.80 | |
| 06:59 | Fix random issue with directory creation when parallel startup causes collisions check-in: 07ee6da7ba user: matt tags: v1.80 | |
| 06:38 | Refactored to move most of the work to scan and recover dead tests out of the servers. This commit also fixes dependencies in the Makefile. check-in: c75b58ded7 user: matt tags: v1.80 | |
|
2023-04-11
| ||
| 21:32 | Throttle writes more than reads using mutex, avoid doing sync when writes are happening using mutex. This is a monte carlo approach. Might need tweaking. check-in: b5df408582 user: matt tags: v1.80 | |
| 18:25 | Switch default cache db to inmem from tmp check-in: aadd589324 user: matt tags: v1.80 | |
| 15:43 | Changed version to 1.8013 check-in: 65629a0fef user: mmgraham tags: v1.80, v1.8013 | |
| 15:12 | Added control over inmem vs. tmp db caching. [setup] dbcache-mode inmem check-in: 9c90eedf66 user: mrwellan tags: v1.80 | |
| 14:23 | added exception handler to deletion of .servinfo file check-in: c96bc5893d user: mmgraham tags: v1.80 | |
| 08:21 | Fixed overlapping calls to sync to on disk db check-in: 4c15e85b4d user: matt tags: v1.80 | |
|
2023-04-10
| ||
| 19:36 | Improved run away server throttling check-in: 4115ca72bc user: matt tags: v1.80 | |
| 16:16 | Some updates from code review check-in: 448f45b91b user: mrwellan tags: v1.80 | |
| 11:58 | Merged fork check-in: 962cf22780 user: mrwellan tags: v1.80 | |
| 07:58 | fixed start up wedging check-in: ce4cc8997a user: matt tags: v1.80 | |
| 06:09 | load control is working but servers are still getting wedged over time check-in: 9771b5d5a9 user: matt tags: v1.80 | |
|
2023-04-09
| ||
| 22:14 | flag loaded at 50 threads. check-in: 4c1e85ecfb user: matt tags: v1.80 | |
| 21:51 | Partial implementation of using the loaded flag to throttle on client side. Need to turn off all server side throttling check-in: b79f855fba user: matt tags: v1.80 | |
| 21:07 | Use run-id in queries looking for test info. Can't assume test-ids are unique. Added separate exception handler for serialization to help with debug (still causes grief when hit but at least you can find the issue but looking in server logs). check-in: ea060a034b user: matt tags: v1.80 | |
| 14:45 | Clean up uses of tcp. tcp6 did not seem to work. Increased tcp backlog (didn't seem to help) and improved backoff check-in: 4860a4e6aa user: matt tags: v1.80 | |
| 13:27 | Added host name to messages about server not started check-in: 21f45d51cf user: mmgraham tags: v1.80 | |
|
2023-04-08
| ||
| 11:48 | Added better fallback on communication failure. check-in: 7dcbd017e3 user: matt tags: v1.80 | |
| 02:45 | Removed assert from simple lock. Removed use of statement cache which fixed bind issues check-in: 532f050f93 user: matt tags: v1.80 | |
|
2023-04-07
| ||
| 08:25 | Added switchable support for db on /tmp instead of inmem. Added couple asserts to help find why run-id and servers are not aligned in some cases. check-in: cfcc13973c user: matt tags: v1.80 | |
| 03:56 | For /tmp db cache disambiguate the db with pid check-in: 481acc5191 user: matt tags: v1.80 | |
|
2023-04-06
| ||
| 22:32 | Moved few functions around. Added beginnings of being able to use /tmp/ db in place of inmem check-in: 5e665a1f75 user: matt tags: v1.80 | |
| 20:25 | Cherrypicked fix for ticket c10775f9d83a4e29f50f9ccdbfc9fd326f493e2e from e37f check-in: 8982b550f7 user: matt tags: v1.80 | |
| 20:22 | Cherrypick forward 07ba2e and 3e5f, chicken 5 work Leaf check-in: 3d39f8f19d user: matt tags: v1.80-chicken5-stuff | |
| 20:00 | cleanup - 57d44 check-in: 5dac6d2e49 user: matt tags: v1.80 | |
| 19:26 | Some cleanup and more added to plot units as graph (c1077 and 198ba). check-in: 2336d19a47 user: matt tags: v1.80 | |
| 19:19 | Bunch of cleanup. Ran pretty well, no worse than last commit and maybe bit better. Got as far as y/b/a and w/b/a check-in: 7bfbd68003 user: matt tags: v1.80 | |
| 16:37 | merge-fork check-in: c574c7b21b user: matt tags: v1.80 | |
| 16:25 | Added asserts, improved message/assert on lock failure check-in: eca566d9f9 user: mmgraham tags: v1.80 | |
| 13:32 | Make server logs not overlap on server.log - this makes debug easier check-in: 3053005860 user: matt tags: v1.80 | |
| 09:23 | Fixed performance issue with append in runs.scm check-in: 0dc6c83d6e user: matt tags: v1.80 | |
| 08:01 | Fixed MTLOWEST load check-in: 71f876d389 user: matt tags: v1.80 | |
| 06:35 | Turn on portlogger check-in: 5cf2865c15 user: matt tags: v1.80 | |
| 06:25 | Return to using portlogger for getting candidate port numbers. Got dependencies fixed so make -j now works check-in: c8631e33de user: matt tags: v1.80 | |
| 01:48 | Changed a missed .megatest to .mtdb. Removed useless thread. Removed old commented code. check-in: 909cd71d01 user: matt tags: v1.80 | |
|
2023-04-03
| ||
| 18:41 | Changed tmpdir path to toppath in dashboard:database-changed? to correct run control tab slowness. Removed 2 duplicate functions. check-in: 8a59e78cc4 user: mmgraham tags: v1.80 | |
| 15:22 | Cherrypicked fix for remove-keep check-in: 5bfcab67c7 user: matt tags: v1.80 | |
|
2023-03-27
| ||
| 19:33 | Changed version to 1.8012. check-in: bcc22ebf3c user: mmgraham tags: v1.80, v1.8012 | |
| 18:15 | Lots of proper use of debugprint and mtargs plus some small cleanup. check-in: 7060a6d776 user: mrwellan tags: v1.80 | |
| 15:51 | merged fork check-in: f7059c9604 user: mrwellan tags: v1.80 | |
| 15:49 | Removed use of margs check-in: fba5bad548 user: mrwellan tags: v1.80 | |
| 13:06 | changed more cases of .megatest to .mtdb check-in: 85ebde8764 user: mmgraham tags: v1.80 | |
|
2023-03-22
| ||
| 19:40 | Changed .megatest to .mtdb in several places check-in: 332dd9df31 user: mmgraham tags: v1.80 | |
| 13:01 | Merged fork check-in: 9fe68a8fb2 user: mrwellan tags: v1.80, v1.8011 | |
| 09:05 | Merged mtdb change. check-in: a50d34f842 user: mrwellan tags: v1.80 | |
|
2023-03-21
| ||
| 21:09 | Converted .megatest to .mtdb Leaf check-in: f00c7b0bd9 user: matt tags: v1.80-mtdb | |
| 20:07 | Change .megatest -> .mtdb check-in: 8bd0fd0763 user: mrwellan tags: v1.80-mtdb | |
| 16:10 | disabled syncing of test steps and test data check-in: c2e29b696d user: mmgraham tags: v1.80, v1.8009 | |
| 08:46 | Merged fork check-in: 747321a245 user: mrwellan tags: v1.80 | |
| 08:35 | Merged fork check-in: 664605b88d user: mrwellan tags: v1.80 | |
| 08:22 | Fixed dashboard refresh and test control panel setting of state/status check-in: 35a835818f user: matt tags: v1.80 | |
|
2023-03-20
| ||
| 10:17 | changed version to 1.8009. check-in: 901bbd55b3 user: icfadm tags: v1.80, v1.8009 | |
| 07:31 | Make nfs mode default for dashboard check-in: aaa4fca591 user: matt tags: v1.80 | |
| 07:09 | Added run-id to test output on import-sexpr for sanity checking by user check-in: 78ea3ef6e3 user: matt tags: v1.80 | |
|
2023-03-19
| ||
| 22:16 | Added -import-sexpr check-in: 996a9b8e3d user: matt tags: v1.80 | |
|
2023-03-17
| ||
| 11:31 | Partial implementation on import sexpr check-in: 7d130344e0 user: matt tags: v1.80 | |
|
2023-03-16
| ||
| 21:18 | Added skeleton of sexpr run importer check-in: dd23dd3b14 user: matt tags: v1.80 | |
| 06:13 | Beginnings of dual ck5/ck4 build support. check-in: fdfdc48e5f user: matt tags: v1.80 | |
|
2023-03-15
| ||
| 10:03 | I saw a couple tcp errors with threads in flight of 500. Reducing to 200 and got clean sixtyfivek fast run. check-in: 34fa77c2e3 user: matt tags: v1.80 | |
|
2023-03-14
| ||
| 21:42 | This combo seemed about as robust as any check-in: 80c20a647d user: matt tags: v1.80 | |
| 11:17 | Merged fork check-in: d8ef55d077 user: mrwellan tags: v1.80 | |
| 09:26 | Fixed couple issues with setting server timeout. Reverted viewscreen to start in background as it didn't work properly without that. check-in: dda2fe1e9e user: matt tags: v1.80 | |
| 08:06 | Make server timeout configurable using existing settings check-in: ad08451b0a user: matt tags: v1.80 | |
|
2023-03-13
| ||
| 12:16 | Changed version to 1.8008 check-in: d917a7ad37 user: mmgraham tags: v1.80, v1.8008 | |
| 08:37 | Updated transport-mode templates check-in: 83e24c295f user: matt tags: v1.80 | |
|
2023-03-12
| ||
| 12:47 | Got original sync working again check-in: 21debe6bae user: matt tags: v1.80 | |
| 08:54 | Fixed no-sync-db crash check-in: d79897a765 user: matt tags: v1.80 | |
|
2023-03-11
| ||
| 20:47 | Remove noise check-in: 49abec32c1 user: matt tags: v1.80 | |
| 17:12 | This seems to be working but it should not. check-in: ab6d2aae1a user: matt tags: v1.80 | |
| 14:40 | Bit better check-in: 3db414f444 user: matt tags: v1.80 | |
| 13:10 | Partially complete update to configuration' check-in: 0f8bf614e9 user: matt tags: v1.80 | |
| 12:39 | fixed unprotected vector-length check-in: c2a555afb1 user: matt tags: v1.80 | |
| 10:37 | WIP, getting nfs working again check-in: 79f9af8364 user: matt tags: v1.80 | |
|
2023-03-03
| ||
| 18:32 | no crashes (yet) check-in: 349234dd61 user: mrwellan tags: v1.80 | |
| 17:37 | wip check-in: abb2c73974 user: mrwellan tags: v1.80 | |
| 02:16 | Speculative fix for wrong server access issue check-in: 0f348aa84b user: mrwellan tags: v1.80 | |
|
2023-03-02
| ||
| 21:32 | Improve some of the logic around servers check-in: 87813c75d2 user: mrwellan tags: v1.80 | |
| 19:19 | Better message check-in: c3879943f0 user: mrwellan tags: v1.80 | |
| 19:12 | Locking of db check-in: 494cb9b035 user: mrwellan tags: v1.80 | |
|
2023-03-01
| ||
| 21:29 | Speculative fix for transaction crash. Check if there is a transaction before proceeding. check-in: e8d8a38e49 user: mrwellan tags: v1.80 | |
| 10:22 | Added template for dashboard-transport-mode.scm check-in: 50f20490df user: mrwellan tags: v1.80 | |
| 07:41 | ATTACH sync working ok. check-in: f803cb519f user: matt tags: v1.80 | |
|
2023-02-28
| ||
| 20:27 | ATTACH Sync not working. check-in: 692ebd2f40 user: mrwellan tags: v1.80 | |
|
2023-02-27
| ||
| 21:25 | Added function for using ATTACH for sync between db's. Not yet in use. check-in: dca3a45c98 user: mrwellan tags: v1.80 | |
| 10:54 | Added suicide mode to db:with-db for development. It has been very hard to isolate threaded crashes under the tcp-server. check-in: 3b54f22608 user: mrwellan tags: v1.80 | |
|
2023-02-23
| ||
| 11:53 | Fixed get-target check-in: 78fc9c5443 user: matt tags: v1.80 | |
| 08:46 | configf ext-tests fixed. check-in: 5cb7bf9076 user: matt tags: v1.80 | |
| 06:04 | rmt:get-key-val-pairs was using wrong db. check-in: 335653cb5f user: matt tags: v1.80 | |
|
2023-02-22
| ||
| 12:27 | Improved messages on -list-runs check-in: c8040a78fa user: matt tags: v1.80 | |
| 11:42 | bumped allowed number of running servers check-in: 5fca83553e user: mrwellan tags: v1.80 | |
| 11:12 | Merged tcp-inmem check-in: 47a4f89869 user: mrwellan tags: v1.80 | |
| 11:05 | Oops. Was not creating .megatest directory. Leaf check-in: 0386246065 user: mrwellan tags: v1.80-tcp-inmem | |
|
2023-02-21
| ||
| 21:35 | Silence noise from tcp-server. Turn on throttling. check-in: a8b8aa106f user: matt tags: v1.80-tcp-inmem | |
| 20:25 | Fixed few minor issues and got creation of temporary no-sync in .megatest dir working ok. check-in: 6e2393b843 user: matt tags: v1.80-tcp-inmem | |
| 17:58 | Switching default to tcp/inmem on build check-in: 2b61f8385c user: matt tags: v1.80-tcp-inmem | |
| 17:47 | Fixed sync back check-in: 5e7e64a893 user: matt tags: v1.80-tcp-inmem | |
| 17:02 | merged fork check-in: d8fae05b29 user: matt tags: v1.80-tcp-inmem | |
| 16:54 | Fixed run-id issue that caused wrong db to be addressed. check-in: c2f5ef0caf user: matt tags: v1.80-tcp-inmem | |
| 15:40 | wip check-in: b51df9a459 user: matt tags: v1.80-tcp-inmem | |
| 11:39 | server start smooth, but initial data load to inmem broken. check-in: 677b6ef8e8 user: matt tags: v1.80-tcp-inmem | |
|
2023-02-20
| ||
| 20:05 | Hacky solution for no-sync db check-in: b8f2578046 user: matt tags: v1.80-tcp-inmem | |
| 14:17 | wip. better but now there are run-away issues in ext-tests: check-in: 94e5d1fb43 user: matt tags: v1.80-tcp-inmem | |
| 10:32 | wip check-in: 0ac0c8e72f user: matt tags: v1.80-tcp-inmem | |
| 08:35 | wip, dashboard and list-runs work check-in: 4eb82b3919 user: matt tags: v1.80-tcp-inmem | |