3.6.6

[check_openmanage.git] / CHANGES

3.7.0
------------------

* Major overhaul of the perfdata code. This includes API breakage and
  fixes the following:
  - Probes were not sorted correctly
  - Voltage data was not included
* The PNP template check_openmanage.php has been redone to work with
  the changes in perfdata output from the plugin
* Added support for a configuration file

3.6.6   2011-04-28
------------------

* Fixed typo in help output
* SD card check is now included if the parameter '--only chassis' is
  specified
* The plugin will issue a proper warning if a physical drive is
  uncertified, instead of an unspecified warning. One or more
  uncertified drives will make the controller go into a non-critical
  (warning) state.
* Slightly improved reporting of fan status
* Exit with value 3 (unknown) if printing debug, help or version
  info. This is considered best practice for Nagios plugins.
* Workaround added for logical SAS connectors to external storage
  enclosures, when using check_openmanage in local mode with OMSA
  6.4.0 or later versions. The output from omreport could contain
  lines that the plugin was unable to parse, which would lead to
  internal errors.

3.6.5   2011-02-09
------------------

* Fix counting of components when blacklisting is used. Components
  should be counted even if blacklisted
* Added some unsupported vdisk types to the list. The OMSA MIB
  identifies these, but lists them as unsupported.
* Added option '-B' or '--show-blacklist' to show any blacklistings in
  the OK output
* Fixed a bug for checking voltage probes, if the reading is missing
  via SNMP.
* Fixed a regression bug for a power monitoring corner case.

3.6.4   2011-01-04
------------------

* Added more robustness wrt. values from OMSA obtained via SNMP, to
  avoid internal errors where non-important values are missing.

3.6.3   2010-12-13
------------------

* A few compatibility fixes for OMSA 6.4.0 were added.

3.6.2   2010-11-25
------------------

* Added support for IPv6 when checking via SNMP. IPv6 can be turned on
  with the option '-6' or '--ipv6'. The default is IPv4 if the option
  is not present.
* Added support for TCP when checking vis SNMP. The option '--tcp' can
  be used to turn on TCP. The default transport protocol is UDP if the
  option is not present.
* The mode of operation (local or SNMP) is shown in the debug
  output. If SNMP is used, the debug output will also show the SNMP
  protocol version, IP version and transport protocol (UDP or TCP).
* Amperage probe status via SNMP is of type "probe status", not
  regular status. This has been fixed.
* Massive overall robustness improvements to handle OMSA bugs where
  some information from OMSA is missing.
* Memory module enumeration via SNMP changed somewhat to reflect
  enumeration provided by omreport. This ensures that the plugin's
  output is identical in SNMP or local mode wrt. dimms IDs.
* Fan enumeration via SNMP changed somewhat to reflect enumeration
  provided by omreport. This ensures that the plugin's output is
  identical in SNMP or local mode wrt. fan IDs.

3.6.1   2010-11-02
------------------

* Included new check for SD cards. Newer servers such as the R710 can
  have SD cards installed, these should be monitored. The SD card
  check is on by default. A new blacklisting keyword 'sd' has been
  added. The SD card check can be turned off with '--check sdcard=0'.
* Handle special cases where power monitoring capability is disabled
  due to non-redundant and/or non-instrumented power supplies.
* For physical disks probed via SNMP, check that values for vendor,
  product ID and capacity is available before attempting to display
  those values.
* If a physical disk is in sufficiently bad condition, the vendor
  field reported by OMSA may be empty. The plugin now handles this
  situation without throwing an internal error.


3.6.0   2010-08-30
------------------

* Storage is no longer allowed to be absent. If the plugin doesn't
  find a storage controller, it will give an alert. For diskless
  systems or servers without a Dell controller that OMSA recognizes
  you will now have to specify '--no-storage' or '--check storage=0'
  to work around this.
* Report the system revision (if applicable) wherever the model name
  is printed. E.g. "PowerEdge 2950 III" instead of "PowerEdge 2950".
* Small change in search path for omreport: The new location for OMSA
  6.2.0 and later on Linux will be attempted first.
* Small bugfix for the '--check' parameter, if the argument is a
  filename. The file could not contain a linebreak, this has been
  fixed.
* Added an option '--no-storage', which is equivalent to the general
  option '--check storage=0'.


3.5.10  2010-07-14
------------------

* If a physical disk is a hot spare, display this information in
  the debug output
* Report the bus protocol (e.g. SAS, SATA) and media type (e.g. HDD,
  SDD) for physical disks in the debug output, if applicable
* Minor fix for 100GB physical disks, write "100GB" instead of "99GB"
* SNMP: Use new features of OMSA 6.3.0 to display occupied and total
  slots in storage enclosures, if applicable. This information is not
  available with omreport and check_openmanage will not display this
  info in local mode.
* SNMP: Added new processor IDs from the OMSA 6.3.0 MIBs
* SNMP: Use connection tables in a proper way to determine controller
  and enclosure IDs, for use with physical disks and enclosure
  components (fan, temp sensors etc.). This fixes a long standing bug
  for servers with more than one controller, if checked via SNMP.
* SNMP: Use the nexus ID as last resort to find the controller for
  physical disks. Workaround for older, broken OMSA versions.
* SNMP: Identify enclosures (e.g. '2:0:0') properly so that the
  reporting with SNMP corresponds to the same report with omreport.
* SNMP: added a couple of workarounds for pre-historic OMSA versions


3.5.9   2010-06-29
------------------

* More fine-grained reporting of temperature warnings for enclosure
  temperature probes.
* Max/min temperature limits for enclosure temp probes are reported in
  the debug output
* Report enclosure temperature probes that are "Inactive" as ok
* Don't try to print out the reading of enclosure temperature probes
  if the reading doesn't exist or is not an integer
* Report enclosure EMMs that are "Not Installed" as ok, instead of
  critical
* Corrected typo in the PNP4Nagios template


3.5.8   2010-06-17
------------------

* Remove reporting of which controller a logical drive is "attached"
  to, since this information can't be reliably extracted via SNMP.
* Avoid collecting Lun ID via SNMP for virtual disks, we don't use it
* Report total memory and number of dimms in the ok output
* Difference in reporting if amperage probes have discrete readings
* Workaround for broken amperage probes
* Added workaround for bad temperature probes that yields no reading
  in SNMP mode
* Get OMSA version via SNMP slightly more efficiently


3.5.7   2010-03-19
------------------

* Added robustness for received SNMP values that are not defined in
  the MIB. Instead of throwing a perl warning when this happens, the
  plugin will not report the undefined value.
* Defined "Replacing" as a defined state for physical disks in SNMP
  mode, even though this state is not defined in the MIB. It is
  reported as such by omreport.
* Physical disk brand/model is now reported when the state of the disk
  is "Rebuilding" or "Replacing".
* The state of a physical disk is reported in parentheses when
  predictive failure is detected. It is useful to know if a disk is
  online, offline, spare or even failed when predictive failure is
  reported.
* Handling of physical disk predictive failure has been improved
  overall.
* Refactoring of the perfdata code. In conformance with the plugin
  development guidelines, the UOM (unit of measure) previously
  reported in the perfdata output has been removed.
* The -p or --perfdata option now
  takes an optional agrument 'minimal', which triggers shorter names
  for the perfomance data labels. This shortens the output and is a
  workaround for systems where the amount of output exceeds the 1024
  char limit of NRPE.
* The PNP4Nagios template has been updated. Users of check_openmanage
  and PNP4Nagios are advised to upgrade. This version of
  check_openmanage needs the new template.
* Lots of other small improvements and updates.


3.5.6   2010-02-23
------------------

* New option '--use-get_table' is added as a workaround for SNMPv3 on
  Windows using net-snmp. This option will make check_openmanage use
  the Net::SNMP function get_table() instead of get_entries() to
  collect information via SNMP.
* Include a blacklisting option 'ctrl_pdisk' which takes the
  controller number as argument. This blacklisting option only works
  with omreport and is a workaround for broken disk firmwares that
  contain illegal XML characters. These characters makes openmanage
  barf and exit with an error. Patch from Bas Couwenberg.
* If the blacklisting keyword "all" is supplied for a component type,
  that component type is not checked at all, i.e. the commands are
  never executed. This will make check_openmanage execute faster if
  blacklisting is heavily used.
* Option '--htmlinfo' now has a shorter equivalent '-I'
* The option '--short-state' now has a shorter equivalent '-S'


3.5.5   2010-01-22
------------------

* Fixed an SNMP bug where the plugin didn't handle OID indexes that
  were not sequential. Thanks to Gianluca Varenni for reporting.
* Fixed an SNMP bug when checking old hardware such as the PE 2650 and
  PE 750. The controller id for physical drives were collected and
  displayed incorrectly. This release uses an additional OID to fetch
  this info, which would otherwise be unavailable. Thanks to Gianluca
  Varenni for reporting this bug.
* Should use %snmp_probestatus, not %snmp_status when checking the
  status of voltage probes. Thanks to Ken McKinlay for a patch.
* Fix when identifying blades via SNMP with very old OMSA
  versions. Patch from Ken McKinlay.
* Better way of finding the ID of physical drives via SNMP


3.5.4   2010-01-13
------------------

* Added support for storport driver version for controllers, only
  applicable on servers running Windows.  A new blacklisting keyword
  for suppressing storport driver messages was added.
* The "all" keyword in blacklisting is now case insensitive.
* More fine-grained reporting in the rare case where a controller
  battery fails during learning and charging states.
* New improved way of reporting perl warnings during execution of the
  plugin.


3.5.3   2009-12-17
------------------

* Fix for path to omreport on Linux with OMSA 6.2.0
* A couple of other small fixes


3.5.2   2009-11-17
------------------

* Fix for undefined device name for logical drives (thanks to Pontus
  Fuchs for a patch)
* Fixed a bug in the PNP4Nagios template, that prevented the template
  from working with PNP4Nagios 0.6. Thanks to the PNP4Nagios team for
  the fix.
* Other small fixes

3.5.1   2009-10-22
------------------

* CPU type, family etc. are now reported in case of a CPU failure (and
  in the debug output)
* The debug output now reports Openmanage version and plugin version


3.5.0   2009-10-13
------------------

* New option '-a' or '--all' turns on checking of everything
* The manual page (POD info) is removed from the script and is now in
  a separate file, to make check_openmanage fully ePN compatible
* ePN is no longer disabled by default, check_openmanage no longer has
  an opinion on whether it should run under ePN or not
* The '-m' or '--man' option is no longer available
* The option '-v' or '--verbose' is renamed to '-d' or '--debug',
  which makes more sense wrt. its usage
* The '-g' or '--global' option is removed. Checking the global health
  status is now default if applicable
* Checking intrusion detection is now turned on by default
* The obsolete option '--snmp' is removed
* The option '--state' now has a shorter equivalent '-s'
* The basename stuff and options '--only-critical' and
  '--only-warning' are now replaced by an option '--only'
* If plugin is run by Nagios, redirect stderr to stdout
* Added option '--omreport', that lets the user specify the full path
  to the omreport binary
* Added non-8bit-legacy default search paths for omreport.exe for
  Windows boxen
* Minor changes to the plugin output, for consistency
* New blacklisting keyword 'bat_charge' disables warning messages
  related to controller cache battery charging. Thanks to Robert
  Heinzmann for a patch.
* For blacklisting, the component ID kan now be "ALL", in which all
  components of that type is blacklisted.
* Man page is moved to manual section 8


3.4.9   2009-08-07
------------------

* Fixed a bug that could cause errors and weird results when checking
  cooling devices (fans) via SNMP. Thanks to Ken McKinlay for spotting
  this bug and reporting it.

3.4.8   2009-07-31
------------------

* For failed physical drives, check_openmanage will now output the
  drive's vendor, model and size in GB or TB.

3.4.7   2009-07-24
------------------

* The -s|--snmp option was redundant and no longer does anything. SNMP
  is triggered automatically if the -H|--hostname option is
  present. The -s|--snmp option is kept for compatibility, but has no
  effect.

3.4.6   2009-07-07
------------------

* Added support for performance data (temperatures) from attached
  storage enclosures such as the MD1000

3.4.5   2009-06-22
------------------

* Fixed a regression in the --htmlinfo option when it is not supplied
  with an argument

3.4.4   2009-06-22
------------------

* New option --htmlinfo adds clickable HTML links in the plugin's
  output

3.4.3   2009-06-11
------------------

* Fixed a regression bug in CPU and power supply reporting that only
  affects verbose output
* If blacklisting is used, the global health check (via the --global
  option) is now negated. Checking the global health doesn't make
  sense when one or more components is blacklisted. Thanks to Rene
  Beaulieu for reporting this bug
* The PNP4Nagios template is now included in the tarball and zip
  archive

3.4.2   2009-06-03
------------------

* Improved memory error reporting, when using omreport
* Collect performance data from pwrmonitoring (amperage probes) that
  were previously ignored when using omreport

3.4.1   2009-05-25
------------------

* Improved memory error reporting, when using SNMP
* Other small ehnancements

3.4.0   2009-05-25
------------------

* The plugin is now compatible with the Nagios embedded Perl
  interpreter (ePN) in theory. However, the plugin will not not use
  ePN by default. We don't want any "accidents".
* License is now GPLv3, previously only specified as "GPL"
* New options --only-critical and --only-warning. With these options
  the plugin will only print critical or warning alerts, respectively.
* Bugfixes and speed enhancements in the storage section, when
  checking enclosure components via omreport
* The --okinfo option is now less verbose and more to the point
* Lots of code refactoring for readability, maintainability and
  robustness

3.3.2   2009-05-05
------------------

* Fixed a bug in the storage section, when checking controllers. This
  is an obscure bug that only manifests itself in the odd case where a
  server has multiple controllers, and one of the controllers are
  missing some of the OIDs, in which case these OIDs will be missing
  for the other controllers as well. The change is minor and only
  includes using get_table() instead of get_entries() to collect the
  SNMP result. Thanks to Stephan Bovet for reporting this bug.

3.3.1   2009-04-28
------------------

* The --perfdata option can now optionally take an argument
  "multiline", which makes the plugin produce multiline performance
  data output in a Nagios 3.x way. Not really needed, but the plugin
  output is prettier.
* Added comment within the 10 first lines to disable the nagios
  embedded perl (ePN) interpreter by default for Nagios 3.x
* Improvements in the performance data output. Units are now included

3.3.0   2009-04-07
------------------

* Added --global option, which turns on checking of everything. If
  used with SNMP, the global system health status is also probed, to
  protect the user against bugs in the plugin. If used with omreport,
  the overall chassis health is used.
* Support for SNMP version 3
* New check added: esmhealth. This checks the overall health of the
  ESM log, i.e. the fill grade. More than 80% means a warning message
* Fixed alert log reporting to use the same format as for the ESM log
* Output messages are now sorted by severity
* Minor changes in how out-of-date controller firmware/driver is
  reported
* Code refactoring and cleanup

3.2.7   2009-03-29
------------------

* Use "omreport about" to collect OMSA version. Slightly faster than
  "omreport system version". This should give a small speedup in
  certain configurations
* Fixed typo in output when a logical drive is rebuilding. Thanks to
  Andreas Olsson for reporting
* Improved reporting of ESM log content
* Added omreport.sh as alternate omreport path
* Lots of other small fixes and enhancements

Plus: A few changes to make the plugin work with old PowerEdge models
(e.g. 2550, 2450) and/or old OMSA versions (e.g. version 4.5):

* Use the chassisModelName OID to determine if SNMP works (instead of
  BaseboardType)
* No longer require a response when checking baseboard type via
  SNMP. If there is no response, we assume that we're not dealing with
  a blade server

Thanks to Christian McHugh for help with testing and debugging this
stuff

3.2.6   2009-03-05
------------------

* Use 'omreport system operatingsystem' to collect OS info, instead of
  'omreport system' version which is incredibly slow. This should speed
  things up in certain configurations.
* A few speedups, don't collect information that isn't needed
* Man page fixes

3.2.5   2009-02-24
------------------

* New option --linebreak to specify the separator between line in case
  of multiline output
* Added support for 64bit Windows. Thanks to Patrick Hemmen for a patch
* [Patrick Hemmen] Added install.bat for Windows installation
* [Patrick Hemmen] Improvements on install.sh. Will now install in
  /usr/lib64 for x86_64
* RPMs are now architecture dependent, because of different libdir

3.2.4   2009-02-17
------------------

* New option -o|--ok-info to display extra information when everything
  is ok. The plugin can now display storage firmware and driver info,
  DRAC and BMC firmware, and OMSA version
* Support for setting custom minimum temperature thresholds via the
  -c|--critical and -w|--warning options
* Better and more detailed temperature error reporting
* Bugfix in the amperage report (including performance data). The
  plugin now takes into account the correct unit and measurement for
  amperage probes (other than watts)
* New option --port lets the user specify the remote SNMP port number

3.2.3   2009-02-09
------------------

* Regression fix: Use the older Processor Device SNMP OIDs for older
  PowerEdge models, that don't have the new Processor Device Status
  OIDs. Thanks to Nicole Hähnel for reporting this bug.
* Default output (when there are no alerts) now shows RAC firmware,
  BMC firmware, info about controllers and enclosures (firmware,
  driver).

3.2.2   2009-02-03
------------------

* Regression fix: Ignore unoccupied CPU slots with SNMP probing. This
  fixes a bug introduced in versjon 3.2.1, which would output
  something like this if one or more CPU slots were empty: CPU 1 needs
  attention ()

3.2.1   2009-02-03
------------------

* Use Processor Device Status Table OIDs instead of Processor Device
  Table when checking CPUs via SNMP
* Bugfix: don't report throttled CPUs as warnings when checking via
  SNMP (same as for checking locally)

3.2.0   2009-01-27
------------------

* New options --state and --short-state for displaying service state
  along with the alert
* Lots of small fixes for code readability and maintainability

3.1.1   2009-01-12
------------------

* Support for running natively on Windows (using omreport.exe). Thanks
  to Peter Jestico for a patch.
* Support for compiled Windows version, i.e. check_openmanage.exe is
  now a legal script name.
* Exit with error if script basename is illegal/unknown
* Various small fixes

3.1.0   2008-12-26
------------------

* Use of alternate basenames for checking only one class of components
* Added support for checking the ESM log via SNMP
* Code refactoring for robustness and maintainability
* Numerous small fixes and enhancements
* Added install script in distribution tarball

3.0.2   2008-12-20
------------------

* The script no longer aborts if it can't get system information via
  SNMP. Give a warning instead, as this is not a critical error
* Increased robustness when checking controllers

3.0.1   2008-12-11
------------------

* Man page fix in the 'check' section. Thanks to Ansgar Dahlen for
  reporting this.
* Allow invalid command error from 'omreport chassis pwrmonitoring'
* Various small fixes

3.0.0   2008-12-04
------------------

* Use unique IDs for storage components with regard to blacklisting,
  which means that the blacklisting API has changed
* Added checks for storage components: connectors (channels),
  enclosures, enclosure fans, enclosure power supplies, enclosure
  temperature probes and enclosure management modules (EMMs)
* Improved verbose output
* New option -t|--timeout for setting the plugin timeout
* New option -w|--warning for setting custom temperature warning
  thresholds
* New option -c|--critical for setting custom temperature critical
  thresholds
* Option --check can no longer be specified in its short form (-c)
* Code cleanup and improvements

2.1.1   2008-11-24
------------------

* The workaround for the OMSA bug introduced in OMSA 5.5.0 didn't take
  multiple controllers into account. This has been fixed.

2.1.0   2008-11-19
------------------

* New option -i|--info prefixes all alerts with the service tag
* New option -e|--extinfo gives and extra line of output in case of an
  alert (model and service tag)
* New option --postmsg lets the user specify a post message string,
  with info such as model, service tag etc.
* Options -b|--blacklist and -c|--check can now be specified multiple
  times (actually quite useful)

2.0.9   2008-11-17
------------------

* Slightly improved output for alerts on logical drives (vdisks)
* Now shows a rebuilding physical disk as a warning, as this is
  usually accompanied by a degraded vdisk. Previous versions didn't
  show this at all (omreport classifies it as "OK").

2.0.8   2008-11-14
------------------

* Slightly improved output for charging controller batteries

2.0.7   2008-11-12
------------------

* Bugfix for reporting physical drives with predictive failure (both
  via NRPE and SNMP)

2.0.6   2008-10-30
------------------

* Fix bug in option handling (ambiguous options)
* Slightly improved output if checking the storage subsystem is turned
  off
* Don't complain if there are no logical drives. This is OK. Thanks to
  Jamie Henderson for reporting this

2.0.5   2008-10-29
------------------

* Fix bug in SNMP status level table

2.0.4   2008-10-29
------------------

* Added workaround for a BUG introduced in OpenManage 5.5.0. OM
  sometimes adds a newline in the controller driver version name,
  which leads to problems parsing the output. Thanks to Hiren Patel
  for bringing this to my attention.

2.0.3   2008-10-28
------------------

* (snmp) Improved handling of cases where OM is not working properly

2.0.2   2008-10-27
------------------

* Fixed issue where controller number for physical disks can't be
  established via SNMP (now identifies as controller no. -1)

2.0.1   2008-10-23
------------------

* Correctly identifies and reports error condition in which OpenManage
  has stopped working (it happens)

2.0.0   2008-10-23
------------------

* SNMP support
* Same options for checking, blacklisting etc. supported with SNMP
* Same output with SNMP as with NRPE

1.2.1   2008-09-25
------------------

* Collects performance data if the option '-p' or '--perfdata' is
  supplied.
* New blacklisting directives ctrl_fw and ctrl_driver
  added. Suppresses the "special" warning messages concerning outdated
  controller firmware and driver. Useful if you can't or won't
  upgrade.

1.1.2   2008-08-06
------------------

* Fix getting system model and serial number for newer blades

1.1.1   2008-08-06
------------------

* Three new checks added:
  - System battery probes (typical CMOS battery). Newer poweredge
    models have these
  - Power consumption monitoring (if the server supports it)
  - ESM log, with same functionality as the alert log check. Disabled
    by default.

1.1.0   2008-08-04
------------------

* Internal refactoring: use ssv-formatted output from openmanage,
  resulting in slightly faster execution and increased robustness.
* If /usr/bin/omreport doesn't exist, try
  /opt/dell/srvadmin/oma/bin/omreport.
* Allow for no instrumented/redundant power supplies. Needed on
  low-end poweredge models and blades.

1.0.3   2008-07-25
------------------

* Openmanage reports non-critical warning about throttled CPUs on new
  hardware models. Most og us use ondemand CPU frequency scaling (with
  throttled CPUs as a result). This specific non-critical warning (CPU
  Throttled) is ignored from now on.
* Remove superfluous Celcius sign when reporting temperatures.

1.0.2   2008-07-25
------------------

* Accommodate blade systems with no fans or powersupplies, i.e. accept
  errors from omreport when trying to check fans or powersupplies on
  blade servers.
* Accommodate newer hardware with slightly different omreport
  options. Use the newer options if they exist. Not really necessary
  yet, but deprecated options may be removed in future versions of
  Dell OpenManage.

1.0.1   2008-07-18
------------------

* When everything is OK, check_openmanage now outputs the same info as
  Gerhard Lausser's excellent check_hpasm plugin does for HP servers:

  OK - System: 'poweredge 2850', S/N: 'XXXXXXX', ROM: 'A06
  10/03/2006', hardware working fine, 2 logical drives, 4 physical
  drives

1.0.0   2008-07-15
------------------

* Initial release