ID: 5650
Title: Fixed service discovery page for SNMP cluster hosts
Component: WATO
Level: 1
Class: Bug fix
Version: 1.5.0i3
The service discovery page of WATO was showing a warning message
"Service discovery failed for this host: The check type filter function has not been set "
and no discovery could be performed for this host using the GUI
anymore.
Workaround: Use the discovery mode on the command line via <tt>cmk -II [HOSTNAME]</tt>.
ID: 5649
Title: Fixed performance issue caused by too many parse function calls
Component: Core & setup
Level: 1
Class: Bug fix
Version: 1.5.0i3
Fixed a regression that was introduced with 1.5.0i2. The parse functions
were called not only once per Check_MK check, as it is intended. Each
service of a Check_MK check lead to a new call of the parse function logic.
For example in case you have 100 services of a single check type on one host,
the parse logic was called a hundred times instead of a single time.
ID: 5648
Title: omd stop/start: Add new option to execute site actions simultaneously
Component: Site Management
Level: 1
Class: Bug fix
Version: 1.5.0i3
When having a larger number of sites an "omd start" or "omd stop" operation
may take a lot of time since all processes are handles sequentially. To improve
the overall execution time we have now added the optional -p/--parallel arguments
to handle the site actions simultaneously.
This change has mainly been made to fix long running overall startup / stop times
caused when having a lot of sites on a single system. This can be a problem
especially in clustered environments.
ID: 5634
Title: Fixed slowing of CMC stop operation when SNMP checks are running
Component: Core & setup
Level: 2
Class: Bug fix
Version: 1.5.0i3
The Check_MK Microcore may need a long time for completing stop/restart of the
process waiting for currently running operations to finish.
One possible situation was that the Check_MK check helpers currently execute
checks in form of SNMP queries. A common case is that the helpers wait for
the response of a SNMP device which may take up to minutes, depending on the
"SNMP timing settings".
We have now changed the logic that these SNMP queries are terminated which
leads to an immediately ending Check_MK check helper and also a faster ending
Microcore process.
This change is especially important for cluster setups that have a tight
timeout configured for the stop operation of the OMD sites. With this change
the failover situations should be handled a lot faster and more stable.
ID: 5616
Title: ceph_status.pgs: Added new states 'backfilling' and 'backfill_wait' which are treated as WARNING
Component: Checks & agents
Level: 1
Class: New feature
Version: 1.5.0i3
ID: 5612
Title: Monitoring SNMP hosts with SNMP management boards is possible. New services for SNMP management boards are available.
Component: Checks & agents
Level: 1
Class: New feature
Version: 1.5.0i3
For a single SNMP agent running on a device, each OID has a unique
meaning defined in the relevant MIBs. Devices with SNMP management
boards which can be configured in Check_MK, however, can have
multiple SNMP agents relevant to the device. A Check_MK check plugin
therefore needs to be able to distinguish between data sent by the
management board and data sent by the host itself to avoid picking
random data.
Example: snmp_uptime.
Both data sources provide data for this check plugin and it's not
clear which one should be used.
We address this problem as follows:
In the case of multiple data sources which would use the same check
plugin we implement new ones with the same functionality.
Moreover a new check_info attribute 'management_board' is introduced
which can have one of the following four values:
- MGMT_PRECEDENCE
Use management board address/credentials eg. when it's a SNMP host.
Otherwise host's address/credentials are used.
- MGMT_ONLY
Check is only executed for management boards.
- HOST_PRECEDENCE
Use host address/credentials eg. when it's a SNMP HOST
- HOST_ONLY
Check is only executed for real SNMP hosts.
Furthermore these management board specific services have the
prefix 'Management Interface: ' in their service description.
Example: snmp_uptime and mgmt_snmp_uptime.
snmp_uptime is used for SNMP host and mgmt_snmp_uptime is used
for the related SNMP management board. Second one has the
service description 'Management Interface: Uptime'.
Notes for check plugin development:
- Management board specific check types have the check_info
attribute 'management_board' set to 'MGMT_ONLY'. The service
descriptions of these plugins must start with
'Management Interface: '.
- Convention: Prefix 'Management board'-ONLY check plugin names
with 'mgmt_'.
- At the moment the check_info attribute 'management_board' is
only used for SNMP check types.
- The default value of check_info attribute 'management_board'
is 'HOST_PRECEDENCE'.
- MGMT_PRECEDENCE
Use this key if a host may have a management board. If yes
this check type uses the management board as data source,
otherwise the host's one.
- MGMT_ONLY:
Use this key if a host and its SNMP management board provide
data for the same check plugin.
In order to avoid overlapping discoverable services, create
two check plugin types with same functionality. Make use of
include files.
One is used for the host itself, the other one has the additional
service description prefix 'Management Interface: ' and uses the
check_info attribute 'management_board' which is set to MGMT_ONLY.
Example: Have a look at the snmp_uptime and mgmt_snmp_uptime
checks which are a good example for this feature.
- HOST_PRECEDENCE
Always use the SNMP host's credentials regardless whether this host
has a management board or not.
- HOST_ONLY
This check plugin will be used if you want to contact a real
host and not a management board.
ID: 5489
Title: pvecm_status: Fix parsing of agent output that could lead to a crash
Component: Checks & agents
Level: 1
Class: Bug fix
Version: 1.5.0i3