2018-08-11: v5.7

This is a maintenance release that fixes a bug introduced in last release
It also adds a new feature:

    Adapt sysusage and sysusagejqgraph to changes into sar sysstat tool
    since version 11.5.7. As per sysstat commit 8d635e0:
    
        sar: Update "sar -d" output contents
    
        Replace "rd_sec/s" and "wr_sec/s" fields with "rkB/s" and "wkB/s". These
        fields are now expressed in kilobytes instead of sectors. This also make
        them consistent with iostat's output.
    
        Replace "avgrq-sz" field with "areq-sz". This field is now expressed in
        kilobytes instead of sectors and make it consistent with iostat's output.
    
        Rename "avgqu-sz" field to "aqu-sz" to make it consistent with iostat's
        output.

Here are the list of other changes:

    - Fix per network devices storage. Thanks to Barry Kwok for the report.
    - Apply changes from last release to rsysusage.


2018-08-06: v5.6

This release fixes several issues that was reported during the last
year. It also adds some new features:

  * Show timestamp in ISO format.
  * Add documentation about installation from distribution packages.
  * Replace call to ifconfig with ip.
  * Add last run timestamp to title on GD::Graph images.
  * Add "Section: admin" to debian control file.

Here are the full list of changes:

  - Fix typo in plugin sample.
  - Add documentation about installation from distribution packages.
  - Mark sysusagegraph using GD and GD::Graph as deprecated, it will
    be removed in next makor release.
  - Update year in copyright
  - Fix errors in handling dirty and blocked reports. Thanks to
    Barry Kwok for the report.
  - Replace call to ifconfig by ip when ifconfig is not a available,
    ifconfig has been removed in recent distributions like CentOs 7.
  - Show timestamp in ISO format. Thanks to Claus Regelmann for the
    feature request.
  - Add last run timestamp to title on GD::Graph images. Thanks to
    Claus Regelmann for the feature request.
  - Fix unwanted replacement of version number.
  - Add "Section: admin" to debian control file. Thanks to Matt
    Lewandowsky for the report.


2017-02-11: v5.5

This release fixes several issues that was reported during the last
year. It also adds some new features:

  * Add new report on Active/Inactive/Dirty memory.
  * Add report/monitoring of task number blocked waiting for I/O
  * Add report on huges pages utilisation.
  * Add monitoring/report for fan speed reported by sar.
  * Allow report and alarm on device temperature reported by sar.
  * Add GROUP section to be able to classify remote hosts on the index
    page following their group.

Here are the other changes:

  - Remove .pl extension in messages and documentation. Thanks to
    ak2766 for the report.
  - Update documentation about new type: dirty.
  - Update documentation about blocked report type
  - All sensors monitoring have been commented by default in the
    configuration file.
  - Remove extra space on hugepages labels

2015-01-07: v5.4

This release fixes several issues that was reported during the last two years.
It adds some new features:

    * A new -C option to generate reports for all CPU only instead of per CPU.
    * charts are now drawn in separate HTML pages and loaded into an iframe.

Here is the complete list of changes:

	- Use jqplot library by default in crontab example.   
	- Update copyright year.
	- Add multi:no into configuration file to default samples plugins.
	- Fix not working exclusion of mount point from the disk monitoring
	  type. Thanks to Hofmann Gerhard for the patch.
	- Fix sysusagegraph and sysusagejqgraph that fail if "ssh.log" file
	  is left in data repository. Thanks to Frank Crawford fot the patch.
	- Add -C option to sysusagegraph and sysusagejqgraph to generate
	  reports for all CPU only and not per CPU.
	- Fix parseSarOutput() with output of new sar versions.
	- Fix illegal division by zero on I/O workload reports where read and
	  writes were zero for the measurement period. Thanks to Frank Crawford
	  for the patch.
	- Add support of plugins with "multi" type: when one plugin returns
	  list of several monitored values at all. Useful for something like
	  df monitoring -- when it returns all filesystems. Thanks to Anton
	  Fedorov for the patch.
	- day/week/month/year charts are now draw in separate HTML pages and
	  loaded into an iframe to avoid waiting for data load into a huge
	  single page. Feature requested by Steve Bremer.
	- Change to only used jqplot *.min.js
	- Update jqplot library to version 1.0.3r1118

UPGRADE: just override your current installation backward compatibility is
fully preserved.

2012-04-04: v5.3

	- Extract cursor tooltip from legend with explicit label on all jqplot
	  graphics.
	- Add scaling of the legend values (MAX, CURRENT, AVG). Thanks to Steve
	  Bremer for the feature request.
	- Fix support to device aliases in remote execution. Thanks to Josian
	  Larcheveque for the report.
	- Cleanup some variables initialisation. Thanks to Steve Bremer for the
	  patch.
	- Use a base of 1024 for memory reports. Thanks to Steve Bremer for the
	  patch.
	- Disable storing numbers as integers. Thanls to Steve Bremer for the
	  patch.
	- SysUsage source code is now hosted on github.org. Get it at:
		https://github.com/darold/sysusage
	  Your help to build a better tool is welcome.

UPGRADE: all perl scripts in the bin/ directory have changed so you can just
override your current files.

2012-03-11: v5.2

	- Add current/max/average values on jqplot graphs legend. Thanks to
	  Josian Larcheveque and Steve Bremer for the feature request.
	- The jqplot graph legends has move to the bottom of the graph.
	- Fix detection of an already running sysusagejqgraph instance.
	- Update online documentation to add missing information on cpuall.
	- Update packaging files version.

UPGRADE: only the perl scripts in the bin/ directory have changed so you can
just overide your current files.

2012-03-03: v5.1

	- Fix wrong information on hostname, uptime and sysstat version of a
	  remote server. Thanks to Josian Larcheveque for the patch.
	- Fix missing process name on thread and process reports on remote
	  host. Thanks to Josian Larcheveque for the patch. 
	- Add 'cpuall' target to only report global statistics on all CPUs. Use
	  it if you have lot of CPUs. Thanks to Pavel Niederle for the feature
	  request.
	- Fix an issue in reporting null data from plugins.
	- Add missing alarm notification on plugin threshold exceed. Thanks to
	  Pavel Niederle for the report.
	- Fix missing information into kernel.txt data file for remote hosts.
	- Fix uptime and kernel version report of a remote host.

UPGRADE: only the perl scripts in the bin/ directory have changed. If you want
to replace cpu target by cpuall in your configuration file to only report global
CPU usage, you may want to remove all cpuN and cpudistN data file in the data
repository (just keep files cpu and cpudist, others cpu files with a number can
be removed).

2011-11-24: v5.0

This is a new major version that allow SysUsage to be run in a central place to
collect and graph statistics from multiple remote servers. An other new feature
is the possibility to delay alarm after a given number of attempts.

	- Modify CSS sysusage.css to absolute position instead of relative to
	  fix a display problem on desktop/browser with resolution < 1024.
	  Thanks to Paul Rubin for the patch.
	- Fix positioning of copyright in start page.
	- Change HTML output default install directory to be /var/www/sysusage
	  instead of /var/www/htdocs/sysusage or /var/www/html/sysusage if the
	  htdocs directory doesn't exists.
	- Fix padding between logo and first tab in jqplot output.
	- Remove support to sysstat version < 5.1.3 that do not have support for
	  -p option and migrate the code to get sysstat version from sysusage
	  into Sar.pm
	- Add attempt counter to delay threshold min or max exceed before
	  calling the alarm script. You can specify an attempt number per
	  monitored target by adding an attempt count between brackets after
	  the max and/or min threshold. For example:

		load:8(4):
		proc:httpd:25(3):0(2)

	  this will call the alarm program if the load average is upper than 8
	  after 4 consecutive sysusage run interval. The second will warn if
	  the number of httpd process is upper than 25 after 3 successive run
	  or two successive run if no httpd process is running. The last can
	  be useful if sysusage run just when httpd is restarted. Thanks to
	  Paul Rubin for the feature request.
	- Nagios OK message is now sent after lower threshold alarm recovery.
	- The old sysusage.cfg file is no more renamed into sysusage.cfg.old
	  after an upgrade install, but instead the new one is created as
	  sysusage.cfg.new

	- Use Proc::Queue to limit child process queue size, this Perl module
	  is now required to run sysusage.
	- Add new configuration section [REMOTE hostname] that allow sysusage to
	  be executed on a remote host. This allow to run sysusage from a central
	  server and sysusagegraph one time on this server, see documentation for
	  more information.
	- Add 'remote' configuration option to PLUGIN section to disable/enable
	  the plugin on remote host execution. Default value is 'yes', if you
	  don't want to execute the plugin on remote host set it to 'no'.
	- Allow space characters between bracket and the section's name in the
	  configuration file.
	- Allow use of ':' or '=' character for directive / value separator into
	  the GENERAL, ALARM, PLUGIN and REMOTE sections. MONITOR section still
	  only use the ':' separator.
	- Several directive have also been added to the GENERAL section:
	  PROC_QSIZE, RESRC_URL, SSH_BIN, SSH_OPTION, SSH_USER and SSH_IDENTITY.
	  See documentation.

UPGRADE: backward compatibility with previous data files have been preserved but
those files will be placed under a subdirectory per hostname. You will need to
replace all Perl scripts, javascript and css files so it should be better to
proceed to a full install. New configuration file will be copied as sysusage.cfg.new


2011-01-17: v4.0

	- Add support to jquery rendered graphs. You just have to replace
	  sysusagegraph by sysusagejqgraph on you cron command line. With this
	  graph renderer configuration options FLAMING,HIRES,LINE_SIZE are not
	  used. This feature use jqPlot a plotting and charting plugin for the
	  jQuery Javascript framework. See: http://www.jqplot.com/ for more.
	  Thanks to Chris Leonello for his great library.
	- Add missing help/info on some sar output.
	- Fix some CSS errors in sysusage.css
	- Update documentation.

UPGRADE: This release is fully compatible with 3.3 version, so if you don't want
to reinstall all using Makefile.PL you can just copy sysusagejqgraph into the
bin directory and untar resources/jqplot-sysusage.tar.gz into the HTML output
directory and replace the css files of the HTML output directory with the css
files from the resources directory.


2010-07-30: v3.3

	- Add I/O workload per device. I/O workload is the relation between TPS
	  (transfers per second) and IOPS (I/O operations measured in seconds)
	  of a device. If the tps returned by sysstat reach the max theoretical
	  IOPS, your storage subsystem is saturated. See 'dev' monitoring config
	  directive and Documentation/Configuration/Specials Cases
	- Add indication of the maximum I/O throughput of a disk, and may
	  suggest where a I/O bottleneck can occur. This is based on:
	  Maximum disk throughput (KB/s) = (rd_sec/s + wr_sec/s) * 50 / %util
	- Change sysusage graph label "Percent cached swap used" by
	  "Percent swap used by cache".
	- Alarm on high memory load is know calculated in percentage of real
	  memory used, i-e: without cache and buffer memory usage.
	- Suppress install warning on hddtemp and sensors if they not exists.
	- Allow multiple recipient to alarm report by using a coma separated
	  list of email adresses as value of the TO configuration directive.

UPGRADE: Just override bin/sysusage and bin/sysusagegraph.

2010-02-01: v3.2

	- Fix use of wrong 'work' threshold notification value, Kb instead of %.
	- Add submenu in Plugins section. See 'menu' configuration option in
	  [PLUGIN] section.
	- Add new line char after Threshold Max legend in graph.

UPGRADE: Just override bin/sysusage and bin/sysusagegraph.

2010-01-12: v3.1

	- This new release integrate the great work of Marat Dyatko on new
	  web design. I hope you will love it. Thanks a lot Marat !
	  This include Code licensed under the BSD License from Yahoo! Inc.
	  This is to have IE compatibility on CSS menu. If this brake something
	  for you, let me know.
	- Meta-tag refresh has been removed in favor of JavaScript refresh to.
	  only refresh graphs images without full page reload. That's an other
	  addon of Marat Dyatko.
	- Moved css file, Javascript, logo and extra images to resources/ dir.
	  So if you want you can customized more easily the web design.
	- Fix problem with localization. Output pf the sar, and df command are
	  now forced to default locale C. Every command is executed with the
	  environment variable LC_ALL=C.
	- Fix utf8 and russian(all?) locale, added "use open ':locale';" in
	  Sar.pm. Thanks to Ilya for the patch.
	- Fix bug in check_running() method in sysusage where it was reporting
	  a process when none confusing with sysusagegraph. Thanks to Josian
	  Larcheveque for the report.
	- Fix bug on device monitoring where values was always null if device
	  name was containing an underscore. Thanks to Josian Larcheveque for
	  the report.
	- Add missing FLAMING option in default configuration file. Disable
	  by default.
	- Add HIRES configuration directive in GENERAL section to allow fine
	  granularity of the data. Set it to any hour interval (value must be
	  an integer between 1 to 23). Disable by default. Thanks to Michael
	  Bugaev for the feature request.
	- Change process list command 'ps awx ...' by 'ps -e ...'
	- Add 'tproc' configuration directive in MONITOR section to monitor
	  number of thread instead of process (proc). Process are monitored
	  with 'ps -e' and thread with 'ps -eL'.
	- Update packaging script with new sysusage site URL.
	- Set disk usage exclusion as dynamic menu on sysusagegraph to avoid
	  displaying link to old data file.
	- Update web site to reflect changes and add a link to an html page
	  referencing sites hosting SysUsage Plugins.

UPGRADE: Just overwrite the perl Sar.pm module, perl scripts and copy CSS, JS
and images files from the resources/ directory to the HTML output dir. This
release is fully backward compatible with data files.

2009-09-29: v3.0

	- Restrict network report to active interface.
	- Add device aliasing to change device name in web interface
	- Rewrite install script, there's no more question. It is now installed
	  with a default configuration file
	- Makefile.PL now allow command line arguments to overwrite default
	  install path
	- The sysusage.spec file used to build RPM package has been updated and
	  moved to the packaging/RPM/ directory. This directory also include
	  all files to make a Slackware or debian package
	- Add a SlackBuild in packaging/slackware/ to create Slackware package
	- Complete rewrite of the interface with CSS dropdown menu
	- Allow regexp in process monitoring to differentiate some process
	- Remove warning when kernel.txt not exist at first sysusagegraph run
	- Add URL configuration directive in the [ALARM] section to overwrite
	  the default url 'http://hostname.domain/sysusage/' sent into the
	  warn message. Thank to Josian Larcheveque for the report
	- Add [CUSTOM name] section in configuration file to allow the use
	  of monitoring plugins. See README file for more details
	- Add plugin example into the plugin directory
	- Add help at end of graph views to explain what are the graphed values
	- Update documentation to reflect these new features
	- Complete rewrite of CPU monitoring which now show distribution usage
	  (user/nice/system), global usage and virtual cpus usage (steal/guest)
	- The 'wait' configuration keyword is now obsolete. See CPU monitoring
	  rewrite
	- Change PID_FILE configuration option to PID_DIR
	- Add read/write sectors per second for device, automatically added if
	  the "dev" monitoring type is present in config file
	- Add hyperlink to Sebastien Godard sysstat tool into HTML header
	- Fix alarm on I/O request that was overwrite by I/O block
	- Add SKIP configuration directive into ALARM section. There's now two
	  possibility to desactivate sysusage on a given period. If SKIP is set
	  in the GENERAL section sysusage will stop immediatly with no pooling
	  If it is set into the ALARM section, it will not send alarm in the
	  given period. The last one was the old behaviour
	- Update licence information as SysUsage is now under GPL v3
	- Add a daemon mode to run sysusage under the minute so without the
	  help of crond. See DAEMON configuration directive
	- Prevent sysusage to run if INTERVAL is less than 10 seconds and more
	  than 300 seconds (5 minutes);
	- Add average of CPU time for I/O requests issued on device. This is
	  automatically included with 'dev' configuration type
	- Add inode usage with disk space monitoring
	- Add rrdtool graph random flamming style for fun
	- Fix bug in hddtemp report where rrdfiles were not created
	- Add Hrule in rrdtool graph to show thresold max when available
	- Fix case with device with / in sar report.

    UPGRADE: you need a fresh install as many rrdfiles has changed. You may
	rename your html output directory if you want to preserve history
	Version 3.0 is not compatible with older version


2009-07-08: v2.12

	- Fix incorrect warning on socktw report type. Thank to Josian
	Larcheveque for the report
	- Add report of the number of active/passive TCP connections per second
	See output of "sar -n TCP"
	- Add report of the number of receveid/sent TCP segments per second
	See output of "sar -n TCP"

2009-06-11: v2.11

	- Fix memory usage report. Real used memory (without cache) was wrong
	Thank to Stephane Silly for the report
	- Add memory cached+kernel buffer report in memory stats
	- Add report of socket in TIME_WAIT state

2009-02-10: v2.10

	- Fix missing replacement of underscore in pathname for queue or disk
	- Fix missing prompt on "What device do you want to monitor" question
	- Fix warning when hddtemp is not found during installation
	- Force df to only report local disk usage excluding tmpfs
	- Add monitoring of sensors temperature and fan speed following regexp
	  See INSTALL file for more details (chap. SENSORS)

2008-09-15: v2.9

	- Sysusage now support new versions of sysstat > 8
	- Add regexp in disk monitoring exclusion
	- Remove plist-sz and runq-sz from loadaverage report to render only
	  load average for the last, 5 and 15 minutes load
	- Add RRD information to debug mode
	- You can now run debug mode with the --debug parameter with sysusage
	  and sysusagegraph
	- Memory is now returned in bytes to fix Y-axis label (Gb was shown as
	  Mb)
	- Memory max threshold MUST now be given in percent (so lower than 100)
	- Change/clarify some graph label/legend
	- Add CPU usage for device (see dev monitoring configuration option)
	  Following sysstat version use -p option and 'human' device name or
	  you need to give the devm-n for version < 5.1.3. See sar -d 1 1
	- Add amount of memory needed for current workload. This is an estimate
	  of how much RAM/swap is needed to guarantee that there never is out
	  of memory. The max threshold is the percentage of memory needed
	- Fix detection of already running sysusage instance
	- Running 'perl Makefile.PL QUIET' will create default configuration
	  file and default install
	- Fix open file report as systat now only report open file instead of
	  percentage to filemax
	- Add RPM .spec file and a default configuration file with RH like paths
	  for easy RPM build. This is done in respect of the Fedora Core Team
	  work. To build your rpm: rpmbuild -bb --clean sysusage.spec

2008-06-12: v2.8

	- Add hard drive temperature monitoring using hddtemp

2008-01-23: v2.7

	- Allow regexp in disk usage path exclusion
	- Restrict process count to column 5 of command 'ps axw' return
	- Fix bug in swap alarm report
	- Fix Makefile.PL sysusage.cfg warning
	- Add support for Nagios report using submit_check_result
	- Fix skip check when time range is running over the next day
	- Add multiple skip time range support
	- Move project to SourceForge

2007-10-08: v2.6

	- Add support for sysstat > 8.0 with network rx/tx stat now
	reported in Kb

2007-04-30: v2.5

	- Fix locale parsing problem

2007-02-21: v2.4

	- Add CPU iowait statistics
	- Update documentation
	- Add -P option to 'df' system command to prevent line break

2006-12-12: v2.3

	- Add exclusion of mount point from disk space monitoring. Done
	by adding a semi-colon separated list of mount point to exclude
	Ex: disk:98:/oracle/data;/home

2005-07-24: v2.2

	- Add auto-removal of pid file if no sysusage or sysusagegraph
	process are running. Use 'ps -ef' system command

2004-02-16: v2.1

	- Add time range to skip sending alarm

2003-10-10: v2.0

	- Complete code rewrite. Use rrdtool now
	- Improve speed, run hundred time faster
	- Simplify configuration
	- Add alarm report by mail

	See documentation for more details

2003-07-01: v1.1

	- Add Min Threshold on process monitoring

2003-02-06:

	- Initial revision 1.0

