summaryrefslogtreecommitdiffstats
path: root/runtime/queue.c
Commit message (Collapse)AuthorAgeFilesLines
* bugfix: regular workers are no longer started if DA worker is activePavel Levshin2013-11-071-8/+7
| | | | | form mail: Look, below, if it is time to activate DA worker, we call it explicitly. But in this case we do not advise regular workers. They are likely already running at this point, but it is not guaranteed. What if, for example, the system is set to start additional workers when the queue is going over high watermark? What if HighWatermark is set to 1? Regular workers will not be started, and DA worker may fail. Thus, it is reasonable to advise regular workers even if we are going DA.
* cleanup no longer needed codeRainer Gerhards2013-11-071-4/+0
|
* do not adjust batch size for disk queuesRainer Gerhards2013-11-071-1/+1
|
* do not complain for queue size on DA queuesRainer Gerhards2013-11-061-1/+2
|
* Merge branch 'v7-stable'Rainer Gerhards2013-11-051-6/+14
|\
| * improved checking of queue config parameters on startupRainer Gerhards2013-11-051-6/+7
| |
| * emit warning message if queue.size is set dangerously lowRainer Gerhards2013-11-051-0/+7
| |
* | change main/ruleset queue defaults to be more enterprise-likeRainer Gerhards2013-10-041-2/+4
| | | | | | | | | | | | | | new defaults are queue.size 100,000 max workers 2, worker activation after 40,000 msgs are queued, batch size 256. These settings are much more useful for enterprises and will not hurt low-end systems that much. This is part of our re-focus on enterprise needs.
* | experimental: make impstats return delta values where possibleRainer Gerhards2013-09-071-6/+6
| |
* | Merge branch 'v7-stable'Rainer Gerhards2013-07-191-4/+4
|\|
| * debug: improve queue startup debug outputRainer Gerhards2013-07-191-4/+4
| |
* | Merge branch 'v7-stable'Rainer Gerhards2013-07-171-2/+5
|\|
| * regression fix: %d used for long long types in debug outputRainer Gerhards2013-07-171-1/+3
| | | | | | | | | | regression from recent DA queue bug fix set of patches; not present in any released code.
| * cosmetic bugfix: file name buffer was not freed on disk queue destructionRainer Gerhards2013-07-171-1/+2
| | | | | | | | | | This was an extremely small one-time per run memleak, so nothing of concern. However, it bugs under valgrind and similar memory debuggers.
* | Merge branch 'v7-stable'Rainer Gerhards2013-07-161-2/+26
|\|
| * add note on a potential future troublespotRainer Gerhards2013-07-161-1/+8
| |
| * Fixed return state handling in ConsumerDAAndre Lorbach2013-07-161-2/+19
| | | | | | | | | | | | | | | | | | The queue full loop fix added a problem to the queue when rsyslog was shutdown. This problem has been corrected now. Conflicts: runtime/queue.c
* | Merge branch 'v7-stable'Rainer Gerhards2013-07-151-3/+15
|\|
| * bugfix: 100% CPU utilization when DA queue became fullAndre Lorbach2013-07-151-3/+15
| |
* | Merge branch 'v7-stable'Rainer Gerhards2013-07-131-12/+12
|\|
| * nit: use correct function names in new debug instrumentationRainer Gerhards2013-07-131-12/+12
| |
| * Changed and extended debug output in queue/action classes for further analysisAndre Lorbach2013-07-131-6/+6
| |
* | cleanupRainer Gerhards2013-05-161-1/+0
| |
* | enable ability to read existing encrypted queue fileRainer Gerhards2013-05-151-1/+10
| |
* | clean up crypto provider state files on queue file closeRainer Gerhards2013-05-151-1/+0
| |
* | basic queue file encryptionRainer Gerhards2013-05-131-4/+23
| |
* | enable shuffling of crypto parameters down through queue definitionRainer Gerhards2013-05-101-20/+87
|/
* bugfix: segfault on startup if a disk queue was configure without file nameRainer Gerhards2013-05-081-0/+6
| | | | | Now this triggers an error message and the queue is changed to linkedList type.
* cleanupRainer Gerhards2013-04-241-2/+0
|
* Merge branch 'c7' into tmpRainer Gerhards2013-04-091-3/+4
|\ | | | | | | | | | | | | Conflicts: ChangeLog configure.ac runtime/queue.c
| * make imrelp properly terminate on system shutdownRainer Gerhards2013-04-091-1/+4
| | | | | | | | it didn't do so if it was inside a retry loop
| * add output module interface to facilitate cooperative shutdownRainer Gerhards2013-04-081-2/+0
| | | | | | | | | | | | ... in more complex cases (where receiving SIGTTIN is not sufficient). See also: http://blog.gerhards.net/2013/04/rsyslog-output-plugin-wrangling.html
| * Merge branch 'v6-stable' into tmpRainer Gerhards2013-03-201-1/+1
| |\
* | \ Merge branch 'v6-stable'Rainer Gerhards2013-03-121-1/+1
|\ \ \ | | |/ | |/|
| * | Merge branch 'v5-stable' into v6-stableRainer Gerhards2013-03-121-1/+1
| |\ \ | | | | | | | | | | | | | | | | | | | | Conflicts: action.c queue.c (some code moved from action.c --> queue.c)
| | * \ Merge branch 'v5-stable' into v5-betaRainer Gerhards2012-08-221-3/+5
| | |\ \ | | | | | | | | | | | | | | | | | | | | Conflicts: runtime/ruleset.c
| | | * | undo last queue patch - caused a regressionRainer Gerhards2012-08-011-1/+1
| | | | | | | | | | | | | | | | | | | | some more elaborate patch is needed and will be provided
| | | * | bugfix: DA queue fixed handling of bad queue files.Andre Lorbach2012-07-311-1/+1
| | | | |
* | | | | optimize memory layout for much better cache hitsRainer Gerhards2013-01-151-6/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Moave element status out of batch_obj_t because we get a *much* better cache hit ratio this way. Note that this is really a HUGE saving, even if it doesn't look so (both profiler data as well as practical tests indicate that!).
* | | | | Merge branch 'v7-stable-newmsglock' into master-newmsglockRainer Gerhards2012-12-061-2/+41
|\ \ \ \ \ | | |_|_|/ | |/| | | | | | | | | | | | | | | | | | | | | | | Conflicts: runtime/msg.c runtime/queue.c tools/syslogd.c
| * | | | fix missing functionality: ruleset(){} could not specify ruleset queueRainer Gerhards2012-11-301-1/+40
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The "queue.xxx" parameter set was not supported, and legacy ruleset config statements did not work (by intention). The fix introduces the "queue.xxx" parameter set. It has some regression potential, but only for the new functionality. Note that using that interface it is possible to specify duplicate queue file names, which will cause trouble. This will be solved in v7.3, because there is a too-large regression potential for the v7.2 stable branch.
| * | | | silence compiler warningsRainer Gerhards2012-11-301-0/+3
| | | | | | | | | | | | | | | | | | | | | | | | | the changes do not affect actual code execution, just keep the compile log clean.
| * | | | silence compiler warnings & some cleanupRainer Gerhards2012-11-141-1/+1
| | | | |
* | | | | queue: file delete stream does no longer do real ioRainer Gerhards2012-11-191-62/+34
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This stream is primarily used for state tracking, and has been modified to do just that. This results in considerable less io being done and the respective speedup.
* | | | | queue: reduce CPU load for deserializing message propertiesRainer Gerhards2012-11-061-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | | Linear runtime due to message order. Was quadratic before. However, not a big overall improvement.
* | | | | queue: mini-improvement in deserializer (stage work)Rainer Gerhards2012-11-051-2/+2
| | | | |
* | | | | queue: handle unknown queue type in debug outputRainer Gerhards2012-11-031-0/+3
| | | | | | | | | | | | | | | | | | | | note: can not happen, but...
* | | | | queue: remove time() calls from msg deserializationRainer Gerhards2012-11-031-4/+16
| | | | |
* | | | | queue: use specific deserializer for msg objectRainer Gerhards2012-11-031-3/+2
| | | | | | | | | | | | | | | | | | | | spares lengthy table lookups
* | | | | queue: remove unnecessary (obj_t*) redirection from msg ptrsRainer Gerhards2012-11-031-42/+40
| | | | |
generated by cgit v1.2.3 (git 2.25.1) at 2025-07-15 02:21:27 +0000