1
0
Fork 0
Commit graph

230 commits

Author SHA1 Message Date
Michael Vogel a47ae788ba Improved logging 2016-10-11 06:30:53 +00:00
Michael Vogel ef6a23caf3 Worker: Reserve some space for low priority processes 2016-10-10 21:01:36 +00:00
Michael Vogel 6abac720e2 Small performance tweaks for "item_store" as well. 2016-10-07 06:05:43 +00:00
Michael Vogel 6d16c39b7e dbstructure now switches in the maintenance mode when updating 2016-10-02 13:52:52 +00:00
Michael Vogel 1caa87dce6 Restructured the way, processes are inserted and removed 2016-09-09 20:55:49 +00:00
Michael Vogel 22f32d9721 New process table for a better detection of running workers 2016-09-09 20:33:54 +00:00
Michael Vogel 108a1c4ac6 New priority level for stale processes. 2016-08-18 12:33:17 +02:00
Michael Vogel 67b16507f1 Performance improvements for the poller 2016-08-14 21:02:29 +02:00
Michael Vogel d138b11868 Use a 10 minute timeout for high priority processes. This may be better. 2016-08-08 21:28:44 +02:00
Michael Vogel 29168de677 Process timeouts are now priority depending 2016-08-08 19:20:40 +02:00
Michael Vogel 0411eb289e Lower the priority for stale processes 2016-08-04 15:41:32 +02:00
Michael Vogel 074ae59f49 Reschedule killed processes at the beginning of the queue. 2016-08-04 15:33:15 +02:00
Michael Vogel 7157ca3809 Fastlane is now working with any priority level 2016-08-03 15:59:25 +02:00
Michael Vogel ff430640c0 Small variable type fixed, removed unused includes 2016-08-03 10:19:46 +02:00
Michael Vogel d673f44c5b Split cron jobs in cronjobs, introduce fastlane for high priority tasks 2016-08-03 10:03:05 +02:00
Michael Vogel b9dbb0ace1 Split cronhook call to several single calls 2016-08-02 06:28:34 +02:00
Michael Vogel 668da905e2 "proc_run" is now called with priority. 2016-08-01 07:48:43 +02:00
Michael Vogel 5e81c105e0 We now work with the "priority" field - that's better 2016-07-24 20:25:11 +02:00
Michael Vogel 4be7e5a581 "Fast lane" mechanism to reserve one queue for priority processes 2016-07-24 11:22:44 +02:00
Michael Vogel 7580c8ed11 Processing of high priority processes 2016-07-23 22:57:22 +02:00
Michael Vogel fb447e84ee Some more information about the current queue status of the worker 2016-07-22 18:07:04 +02:00
Michael Vogel 4b29937a74 Worker: New value for cooldown time after each function call 2016-07-13 19:43:16 +02:00
Tobias Diekershoff 598a9f76a6 Merge pull request #2658 from annando/1607-maxload
Central Backend/Frontend detection when checking for overload
2016-07-02 21:52:48 +02:00
Michael Vogel e4c6cfd2d8 Moves the function "is_result" from dba to dbm. 2016-07-02 14:00:42 +02:00
Michael Vogel 7d531afd00 Merge branch 'rewrites/is_filled_array_introduced' of git://github.com/Quix0r/friendica into Quix0r-rewrites/is_filled_array_introduced
Conflicts:
	boot.php
2016-07-02 13:38:23 +02:00
Michael Vogel c549ae939c Added the new function to the poller. 2016-07-02 09:31:28 +02:00
Michael Vogel 0518f004c2 removed duplicated functions 2016-06-04 12:20:09 +02:00
Michael Vogel b45772ba6e Configurable amount of maximum processes 2016-06-04 11:04:26 +02:00
Michael Vogel df337e57d1 Quit when processlist is too long 2016-06-03 23:10:23 +02:00
Michael Vogel 85a9c7d96e The poller now has the logging for processes as well. 2016-06-01 21:54:02 +02:00
Michael Vogel 76b24f61a9 New (experimental) value to define the maximum level of database connections for the worker 2016-04-23 10:11:09 +02:00
Roland Häder 9cfc249b12
Moved is_filled_array() to both dba classes and named it is_result().
Please see ticket #2390 for full discussion.

Signed-off-by: Roland Haeder <roland@mxchange.org>
2016-03-12 18:54:55 +01:00
Roland Häder c60605ce0c
Merge branch 'develop' of github.com:friendica/friendica into rhaeder-develop 2016-03-12 18:50:14 +01:00
Michael Vogel beb2346cfc The function to check for maxload and the lockfile is centralized 2016-03-08 20:28:09 +01:00
Roland Häder d88606321d
Just easier code ...
Signed-off-by: Roland Haeder <roland@mxchange.org>
2016-03-04 22:39:08 +01:00
Roland Häder d5c1ce490b
No processing if error or empty array
Signed-off-by: Roland Haeder <roland@mxchange.org>
2016-03-04 22:38:18 +01:00
Michael Vogel acb09d3a3d Database connections: When we now check for user values we check the system values as well 2016-02-12 11:04:25 +01:00
Michael Vogel 5027fbf733 Poller: Now the user limits can be detected automatically as well 2016-02-11 21:39:34 +01:00
Michael Vogel 4af77b60ce Poller: Use the processlist when the number of maximum database connections was provided manually 2016-02-11 11:33:45 +01:00
Michael Vogel 211be5bfc5 Poller: The maximum number of connections can now be configured 2016-02-09 23:28:33 +01:00
Michael Vogel 7af3dd01d8 Poller: Check the number of used database connections 2016-02-09 06:42:00 +01:00
Michael Vogel 275f7b2660 The worker now kills processes that run longer than 3 hours 2016-01-24 02:53:21 +01:00
rabuzarus 3df5c17124 reformat todo according to doxygen style 2015-12-25 23:17:34 +01:00
Michael Vogel 33f354a68c Bugfix for the maximum load check in worker. 2015-12-16 00:14:53 +01:00
Michael Vogel 6e03477598 Sometimes the function "sys_getloadavg" doesn't return an array. This is a workaround. 2015-12-15 23:26:58 +01:00
Michael Vogel fbbba6828b Some better logging 2015-12-06 20:04:33 +01:00
Michael Vogel 9f8da37c99 Move the process check at the beginning of the script 2015-12-06 16:40:31 +01:00
Michael Vogel 54085508e5 Double check for maximum number of workers 2015-12-06 16:28:28 +01:00
Michael Vogel c3ee255b0c Reduction of double executions - hopefully 2015-09-29 06:15:26 +02:00
Michael Vogel ed8cdc7d85 Logging message added 2015-09-28 22:32:56 +02:00