Activity for Ed Freesmeyer
-
Ed Freesmeyer committed [r128] on SVN
Big improvements to input-threadpool startup code. Fixed 'mark' msgs not being displayed. I think I found that last issues in spooling (certainly the ones that show up quickly. -- recommitted after disabling debug code
-
Big improvements to input-threadpool startup code. Fixed 'mark' msgs not being displayed. I think I found that last issues in spooling (certainly the ones that show up quickly.
-
Got the spooling working (Destination spooling and worker-threadpool queue-spooling). Also fixed some major issues with the RAM-based memory threadpool queueing. This version appears stable (at least in initial testing).
-
Got it -- finally. '--version' will now show the svn revision number for better version tracking.
-
Trying a different approach
-
testing
-
testing for repitition
-
It works !
-
still testing - I may have it now.
-
testing svn interoperability
-
testing svn revsions in C code.
-
Testing how to include svn revsion numbers
-
Testing svn version number addition.
-
Implemented spooling subsystem. Implemented stateFlags structures for consistency and readabilty (as with Syslogd2). Fixes to memory management and stability.
-
Total re-implementation of the spooling sub-system. Re-designed & re-wrote LogFile and LogSocket. Implemented stateFlags structures for more consistent tracking. Impelemented 'ActiveStaged' ('as*') key values to use in place of configuration-specific ('cf*' values) for better consistency. Other fixes related to stability and memory-management.
-
This is not stable -- it is an interim commit while I test massive code updates.
-
This is not a stable checkin. It is an interim checkin of code as part of a larger capability set. This release introduces CheckSources, CheckDestinations, programmable interrupts and initial connection-spooling and queue-spooling (output threadpools only). This release is also the first with a 4-stage processing model (1: input, 2: parsing and template-verification, 3: output-line prep-and-build, and 4: write-to-device (output) threadpools). Yet to be added are queue-spooling for the stage ...
-
Should now be stable. Fixed issues with startup, logging, and operation since last svn commit. Added ability to use 'ddd:n' syntax to produce numeric default value. This version likely still has sevaral bugs, but it no longer locks up or spontaneously aborts on my laptop. (at least not during the last few days.)
-
Fixes to journal-read code. Cleaned up leftover debugging printf statements throughout several files. Added '--nofork' option to make integration with systemd work with 'notify' startup mode. Updated 'CheckSources' funtion to also reset the journal interface (if active) when it is called. System seems very stable now.
-
Removed some overlooked print statements that do not belong in a backgrounded syslog daemon. Stability was unaffected as stdout will have been closed (before being routed to /dev/null).
-
Stable. Added testFilter tool, tracefilters, throttling, extraction. Updated sscalc tool. Added facility ranges for selector strings. Added ability to use multiple 'f/F/p/P' keys in a single filter-element. Several fixes and updates to filter performance as a result. See documentation at https://syslogd2.sourceforge.io for details and usage.
-
found & fixed some more memory-managemt issues (extraneous 'free' commands)
-
fixed sem_init parameter setting & cleaned up a few stray 'printf' commands and comments.
-
More fixes to filter (and testFilter) code. All filter functions now test for proper operation except the numeric field selection code in the '+' option. Adjustments to Syslog include files to put stand-alone 'debug' statements into '#ifdef DEBUG' brackets. This commit is in preparation for more intrusive changes to memory buffer management.
-
New 'testfilter' tool. New filer capability (extract). Major fixes to filter.[ch] for new functionality as well as for longstanding bugs. Fixes to parsetools.c to correct erroneous behavior when parsing 'ignore' input-options. Added test for 'sem_open' in configure.ac to definitively declare that 'sem_open' functionality exists. Cleaned up un-needed code in Makefile and corrected processing for selected targets. Some fixes to the debug subsystem to correct erroneous displays and to improve o...
-
no operational changes. removed some unnecessary comments and some unnecessary '#define' statements. This commit is just to resync svn and my local development directories.
-
more updates & corrections for stability & functionality. Added code to re-open sockets if they get closed during operation.
-
Added 'journal-reset' function to close & re-open journal when notified that new journal-threads have been added.
-
Updates to the fprintlog include file to reopen files/pipes when CAP_RECONNECT is not defined. Succeeded with pipes. Failed with files because I cannot detect write failures to files that have been deleted. Also added new binary suffix ('n') defined as 'g' suffix without CAP_RECONNECT.
-
Forgot some of the componants of the 'tsuc' code. A big piece of previous commit was stub programs to demo/test the 'socket' destination type.
-
Many fixes and improvements to stability, memory mgmt and command-tool.
-
Major stability fixes and added generic socket output.
-
This seems to work well now. No issues seen in base operations. (spooling not tested).
-
Other than CAP_REBUILD (a new CAP_*), this appears stable. Working on CAP_REBUILD next.
-
fixes to accidentally introduced bugs with last update in filter-code. Re-did final 're-assembly' logic in ApplyFilter to account for deleting an sdString. Fixed some minor parsing issues with journal code (mostly oversights). Adjusted journal's SdStringLength estimate.
-
Still testing. Still finding issues while writing demonstrations. This could be a release candidate. Added LossLess queueing. since last commit. Fixed some initialization and some lock-up issues too. (also some bad code that morphed into the 'SplitMsg' function.
-
Stabalized timer issues, fixed UDP issue (sockets not binding to fd), started working on config display of journal parms. (last update: added CAP_JOURNALTHREADS and CAP_VARBUFFER features & changed generic read-thread based on 'pselect' to one based on 'poll'.
-
development checkpoint working on JOURNALTHREADS & VARBUFFERS.
-
Removed popen from list of utilities to make. It has never worked and is extremely low priority to finish.
-
Fixed bogus error in rsu.c & removed simple* dependencies in Makefile.in
-
Added skipto command, corrected test for noforceprintable and fixed 'include' alias to 'includefile'.
-
Fixes to command-tool & host. Adjusted compile-time parameters based on command-tool stats. Added 'base' configuration. Added queue-stats display & stabalized queue-change/thread-count change code.
-
major fixes. This is a development checkpoint commit to prevent catestrophic loss of work.
-
Forgot to update the version patch-level
-
Fixes to command-tool & host code for various bugs & failures resuling in non-execution of kernel mods to mutex lock issues. Added 'demo' and 'base' versions to Makefile.in. Added CAP_STATS functionality for queue high-water & discards.w
-
many fixes to command-tool & interface. Fixes for various problems found during documentations. Implementation of queue-stats & command-tool.
-
fixed a problem with tailfile input not reading a single file. Added '/' to date delimiters in ParseTime. Re-arranged structures.h to more mirror dbd so I can find stuff while comparing.
-
Coming Soon
-
Coming Soon
-
Split out variables ($,var,TimeString) to separatefiles. Added 'Tag', 'NoTag' options for 'msgString' system variable. Removed lists that used to hold parsed variables in strings in favor of storing the filenames in a string and then reading the files during setup. Removed '$', 'var', 'timestring' as config-file variables in favor of '--defaults varfile = <list>'.
-
IP input and cache support added for input & output connections. Updated processing of system variables to allow multiple independent parameter entries in a single input string. Fixed almost all memory issues (1 small (static) leak remains). Code seems stable and fully functional with all time-related processing imported from Syslogd2.
-
Moved Resolve() to hostLookup.c (and DBD2). Commented out one stray debug printf. Fixed some minor failure points with cache operations. -- Generally just a minor stability update.
-
initialized 'p' pointer in RebuildSelectors
-
updated version.h file and commented out some errant debug statements.
-
Rewrite of RebuildSelectors/BuildSelectors. Updates to affected libraries & functions to transition output code from SetStream to F_STREAM. Fixed more memory leaks & code errors resulting in increased stability & more correct performance
-
CurrentStatus
-
Removed aborted objects, rules & target for 'LogBurst' from Makefile.in.
-
Added explicit copyright notices to all source files.
-
Added multiple deployment scenarios in conjunction with documentation. Increased number of 'extra' facilities from 16 to 32. Alphabetized CAP_* entries for better readability. Removed references to 'CAP_VARBUFFERS until ready to formally execute on this CAP_* entry.
-
Fixed rsp tool that was not properly setting permissions when creating the mkfifo device. Updated ValToString in parsetools with new 'T' context to use abbreviated time units. Update parsedisplay to replace reconfigCheckIntervals with filterCheckIntervals & to display schedules as IntArrayToValString instead of IntArrayToString.
-
Updated readsocket to allow hashtag after continuation if 1st line has a hashtag. Updated filters to fix a memory leak around sdVal allocation. updated syslogparsetools to interpret '.none' as '.~*'. Updated parsers to force schedules to start with 0 if user enters <= 0, Some adjustments in OutputThread found while documenting.
-
Looks like everything now works. Added named-pipe input capability (CAP_PIPESIN). Still need more confidence with spooling/despooling and RebuildSelectors.
-
Forgot to update Makefile.in after creating new listsupport.[ch] files
-
Lots of memory management issue fixes resulting in better operation and far cleaner (and more reliable) shutdown operation. Fixed lots of stuff in kernel-read code and TFRT when reading local-kernel file.
-
newHome
-
newHome
-
Forgot to uncomment stdout in untty
-
Fixes to DRT,TFRT,KRT & checkLine to correct long-standing startup/shutdown crash issues & memory various memory access issues & multi-threading issues. Fixes to debug & parsing code to fix dynamic syslog operation (replace WorkerThread entry point with OutputThread entry point). Valgrind reports are now MUCH better !
-
Minor cleanup -- adjusted command-tool to ignore interrupt routines marked as '-- Reserved --'
-
Fixed mode-setting on Linux input sockets.
-
Allowed relative configuration-include files as relative to ConfigDir. Mostly updates to command-tool (tsuc*) and supporting code (including reader-threadpool initialization code) to try to better handle high-traffic start-up situations. Modifications to all display code to remove settings when underlying CAP_* values are not declared.
-
one more time without the debug statements
-
Minor changes to clean up error reports & insure proper shutdown operations
-
various small fixes to fix minor issues found while writing the Demonstrations documentation.
-
Updates to output-recovery code due to recent algorithmic changes. Fixes major errors with RebuildSelectors
-
Filled in some files inadvertantly omitted from svn
-
Cleaning up the build environment within svn.
-
Many fixes to dbd2. Including memory issues found by valgrind. Resetting release-sequence to 'beta-1' ('b1').
-
Home
-
My first release candidate for Syslogd2.
-
Corrected SVN content & prepping for 1st beta release
-
Removed old documentation from svn in favor of file downloads.
-
Removed old documentation from svn in favor of file-downloads.
-
Fixes for memory issues & fixes to make CheckFilters work properly. Found an error in ApplyFilter that got fixed as well.
-
Fixed lots of issues: shutdown, queue-buffer over-runs, time-parsing, etc. Looks like this iteration may near the last (if not the last) prior to cutting a beta-release image. Have been testing the 'g' model. Need to back through with valgrind & check other models for memory leaks or abborations during operation.
-
memory leaks appear to be fixed. testing for eal. Still need to fix and test spooling.
-
Working on memory leaks. All looks good except for userthreads.... Lots of memory-issue fixes.
-
Lots of bugs found & fixed thanks to valgrind. Got PTY output working also.
-
Fixed multiple output issues that were unmasked when run with IP disabled. Improved speed of CheckDestinations somewhat by stat-checking Linux output sockets before attempting to open them.
-
Fixed/wrote makefile install for systemd users. + documentation update on the 'fix'. Added '--systemd' as cmd-line parameter in service-file for systemd.
-
Fixes for filter-application & a glitch with RebuildSelectors not releasing a mutex-lock. Implemented and documented LogFacility to route traffic via syslog after startup (disable with empty-string or 'none'. LogFacility parameter is either 'facility' or 'facility.priority', but only the 'facility' will be used.
-
Forgot to switch to 'production mode' before committing code.
-
Re-wrote BuildSelectors & RebuildSelectors as well as BuildSources & RebuildSources. Still testing for proper functionality while running on my local Ubuntu laptop.
-
Rewrite of BuildSources & RebuildSources. Lots of other (structural & non-structural) bugs fixed. Terminates reliably now.
-
Intermittant status: revised structures for 'OptSpool' and revisions to BuildSources, RebuildSources, BuildDestiantionSelectors along with parameter simplification for RebuildSources an RebuildSelectors. Still needs revisions to RebuildSelectors, flush system, UserThread, Flush*, Log* (write routines) due to 'OptSpool' & Linux impelementations.
-
CurrentStatus
-
CurrentStatus
-
CurrentStatus
-
Home
1 >
