For RedHat and SuSE Linux.
This release fixes a regression where using -d without a search pattern would fail.
It also generalises the behaviour or -d so you can specify minimum expected line counts using -w and -c, if you're not using a search pattern. In such cases -D is then equivalent to "-d -c 1". You can of course continue to use -d and -D in combination with a search pattern as usual.
I've pushed all code to the git repository, found under the Code tab. Here you can find the latest development version of the plugins. The Files tab will continue to host the "stable" releases.
Added lib search path for BSD systems, and added plugin time-out override option (--timeout).
Bugfix release which fixes incorrect globbing when using --logfile-pattern with log files whose paths contain spaces.
This version adds better Unicode and character set translation support. With --input-enc you can now correctly process files that are encoded in UTF-16 or simply in a different encoding from your locale's default. Likewise, with --output-enc you can change the encoding of the output, in case your Nagios server doesn't like UTF-8. And finally, the --crlf option strips CR characters from the output, so DOS/Windows line endings in the logs don't mess up the output.
Merged patch from Wesley Moore for an --add (-A) option, which makes the plugin combine multiple patterns (specified via -p or -P) using AND instead of the default OR.
Several new features and fixes:
- Takes multiple -p and -n options on the command line
- new option: --negate
- new option: --report-first-only, to output first match instead of last
- new option: --missing, to set a custom exit state if the log is missing
- output change: always returns matched line(s) even if state is OK (--quiet supresses this new behaviour)
- Several portability fixes
- Revised documentation
Fixed some portability issues in the seek file auto-generation code.
This is a bugfix release for Win32 systems only, Unix users need not upgrade.
This update finally supports dynamic log files, i.e. filenames that contain timestamps, indexes, or are otherwise variable. There are three new options, -m, -t and --timestamp, to select the filename you want to monitor based on a bunch of criteria, including just "most recent".
Define the directory and fixed filename component with -l, the variable component with -m, and use -t to refine the final selection if there are still multiple options. Some of the more common date(1) format macro's are supported with -m, and with --timestamp you can make these refer to another time than "now".... read more
What I learned today: you have to put the following line in your /etc/sudoers file if you are trying to run plugins using sudo and they are not producing any output:
It seems 3.5 didn't work with the embedded Perl interpreter, oops. This is (hopefully) fixed now.
The check_log3.pl plugin now supports a new --context (-C) option to return lines of "context" before and/or after a match, à la "grep -C".
Note: if you use -e and modify $parse_out, no context will be returned, only $parse_out. If you want to change the output when using -C, modify @line_buffer instead, this list contains the lines of data up to and including the matched line. You cannot change read-ahead lines in this manner.... read more
The rrd2csv_new.pl script has been updated to properly pad columns with missing data, and support a configurable delimiter character.
The plugin now returns performance data.
I've rewritten the new -a feature as it didn't play nice with -e.
- added -a option to make the script output all matching line instead of just the last one (contributed by Ian Gibbs)
- added -P option for reading the pattern from a file, one per line, OR'ed together
- bug fix: no longer returns bogus data in situations where the amount of disk space used reported by a volume is smaller than the quota space used by a QTree in that volume (it now accounts for deduplication, in other words)
Small update to improve the precision of threshold checking on very large filesystems.
The previous version had a very silly bug in it, sorry about that.
Now outputs fatal errors to Nagios rather than just dying.