Werk #2231: More transparently show errors during service discovery

Checks might run into problems parsing the output of an agent plugin. When this happens during a normal check, you will soon know this as the according service goes UNKNOWN and you will be able to create a crash dump.

Errors in single check plugins during service discovery, however, have been silently ignored unless you used --debug on the command line. This behaviour has been changed in the following way:

cmk -I on the command line

Now for each files plugin a warning is being printed, but the discovery continues. --debug will let the Python exception come through and abort the discovery with a precise error message - just as it used to be.

Check_MK Discovery service

The discovery check now in case of an error during the discovery goes to UNKNOWN and produces a crash dump - just as normal checks do. That way you can easily send us a crash report in case of any error. Previously error would be ignored silently.

Service list in WATO

In WATO the list of a host's services will display a yellow error message if one of the plugins fails. This messages has a link for retrying the discovery without the error message and that way switch back to the old behaviours (temporarliy).

Bulk discovery in WATO

The bulk discovery keeps to the old behaviour of silently ignoring errors but has got a new checkbox for letting a host fail if at least one check plugin of the host runs into an error. The default setting of the checkbox is to ignore errors.

To the list of all Werks