Yvan Arnaud
2018-09-13 10:10:59 UTC
Hi,
I recently upgraded from 1.4.0p34 to 1.5.0p4 and had problem with management board using snmp. I had some crashed checks with ucd_cpu_load, hr_cpu and ucd_mem.
Here some report as example:
Crash Type check
Time 2018-09-13 08:32:51
Operating System sles11sp3
Check_MK Version 1.5.0p4
Edition cre
Core icinga
Python Version 2.7.15 (default, Sep 10 2018, 21:48:49) [GCC 7.2.0]
Exception
KeyError ('ucd_mem')
Traceback
File "/omd/sites/my_site/lib/python/cmk_base/checking.py", line 80, in wrapped_check_func
status, infotexts, long_infotexts, perfdata = check_func(hostname, *args, **kwargs)
File "/omd/sites/my_site/lib/python/cmk_base/checking.py", line 159, in do_check
_do_all_checks_on_host(sources, hostname, ipaddress, only_check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/checking.py", line 234, in _do_all_checks_on_host
sources.enforce_check_plugin_names(only_check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/data_sources/__init__.py", line 230, in enforce_check_plugin_names
source.enforce_check_plugin_names(check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/data_sources/abstract.py", line 314, in enforce_check_plugin_names
check_plugin_names, self._for_mgmt_board)
File "/omd/sites/my_site/lib/python/cmk_base/checks.py", line 843, in filter_by_management_board
host_precedence_tcp, host_only_tcp = _get_categorized_check_plugins(found_check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/checks.py", line 895, in _get_categorized_check_plugins
mgmt_board = get_management_board_precedence(check_plugin_name)
File "/omd/sites/my_site/lib/python/cmk_base/checks.py", line 787, in get_management_board_precedence
mgmt_board = check_info[check_plugin_name]["management_board"]
Local Variables
{'check_plugin_name': 'ucd_mem'}
Python Module Paths /omd/sites/my_site/local/lib/python
/omd/sites/my_site/lib/python
/opt/primitive/lib/python/site-packages
/opt/pcm/lib/python
/omd/sites/my_site/lib/python27.zip
/omd/sites/my_site/lib/python2.7
/omd/sites/my_site/lib/python2.7/plat-linux2
/omd/sites/my_site/lib/python2.7/lib-tk
/omd/sites/my_site/lib/python2.7/lib-old
/omd/sites/my_site/lib/python2.7/lib-dynload
/omd/sites/my_site/lib/python2.7/site-packages
Details
Host my-host
Is Cluster Host No
Check Type mk
Manual Check
No
Uses SNMP
No
Inline-SNMP
No
Check Item
Description Check_MK
Parameters
None
I tried to remove the host and add it again with Wato but the problem persisted.
So, I looked at the checks and found that you duplicated some checks for the management board, like mgmt_snmp_uptime for example.
I tried then to duplicate the faulty checks under local/share/check_mk/checks/. First adding only following line in the check_info section:
'management_board' : HOST_ONLY,
Then I duplicated the checks, renaming them as mgmt_<check_name> and modifying the check_info section like that:
check_info["mgmt_<check_name>"] = {
...
'service_description' : 'Management Interface: <check_description>â,
...
'management_board' : MGMT_ONLY,
}
And it worked again.
Question: did someone already have this problem ?
If this is the right way to fix it, could you please add the modified checks into the next patch ?
Best regards
Yvan Arnaud
I recently upgraded from 1.4.0p34 to 1.5.0p4 and had problem with management board using snmp. I had some crashed checks with ucd_cpu_load, hr_cpu and ucd_mem.
Here some report as example:
Crash Type check
Time 2018-09-13 08:32:51
Operating System sles11sp3
Check_MK Version 1.5.0p4
Edition cre
Core icinga
Python Version 2.7.15 (default, Sep 10 2018, 21:48:49) [GCC 7.2.0]
Exception
KeyError ('ucd_mem')
Traceback
File "/omd/sites/my_site/lib/python/cmk_base/checking.py", line 80, in wrapped_check_func
status, infotexts, long_infotexts, perfdata = check_func(hostname, *args, **kwargs)
File "/omd/sites/my_site/lib/python/cmk_base/checking.py", line 159, in do_check
_do_all_checks_on_host(sources, hostname, ipaddress, only_check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/checking.py", line 234, in _do_all_checks_on_host
sources.enforce_check_plugin_names(only_check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/data_sources/__init__.py", line 230, in enforce_check_plugin_names
source.enforce_check_plugin_names(check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/data_sources/abstract.py", line 314, in enforce_check_plugin_names
check_plugin_names, self._for_mgmt_board)
File "/omd/sites/my_site/lib/python/cmk_base/checks.py", line 843, in filter_by_management_board
host_precedence_tcp, host_only_tcp = _get_categorized_check_plugins(found_check_plugin_names)
File "/omd/sites/my_site/lib/python/cmk_base/checks.py", line 895, in _get_categorized_check_plugins
mgmt_board = get_management_board_precedence(check_plugin_name)
File "/omd/sites/my_site/lib/python/cmk_base/checks.py", line 787, in get_management_board_precedence
mgmt_board = check_info[check_plugin_name]["management_board"]
Local Variables
{'check_plugin_name': 'ucd_mem'}
Python Module Paths /omd/sites/my_site/local/lib/python
/omd/sites/my_site/lib/python
/opt/primitive/lib/python/site-packages
/opt/pcm/lib/python
/omd/sites/my_site/lib/python27.zip
/omd/sites/my_site/lib/python2.7
/omd/sites/my_site/lib/python2.7/plat-linux2
/omd/sites/my_site/lib/python2.7/lib-tk
/omd/sites/my_site/lib/python2.7/lib-old
/omd/sites/my_site/lib/python2.7/lib-dynload
/omd/sites/my_site/lib/python2.7/site-packages
Details
Host my-host
Is Cluster Host No
Check Type mk
Manual Check
No
Uses SNMP
No
Inline-SNMP
No
Check Item
Description Check_MK
Parameters
None
I tried to remove the host and add it again with Wato but the problem persisted.
So, I looked at the checks and found that you duplicated some checks for the management board, like mgmt_snmp_uptime for example.
I tried then to duplicate the faulty checks under local/share/check_mk/checks/. First adding only following line in the check_info section:
'management_board' : HOST_ONLY,
Then I duplicated the checks, renaming them as mgmt_<check_name> and modifying the check_info section like that:
check_info["mgmt_<check_name>"] = {
...
'service_description' : 'Management Interface: <check_description>â,
...
'management_board' : MGMT_ONLY,
}
And it worked again.
Question: did someone already have this problem ?
If this is the right way to fix it, could you please add the modified checks into the next patch ?
Best regards
Yvan Arnaud
