Avoid crash on failure to parse int in emerson_temp discovery.

[indigoparadox]

Thank you for your interest in contributing to Checkmk!
Consider looking into Readme regarding process details.

General information

While performing a discovery on certain Emerson/Vertiv devices, they can return an SNMP value to the generic Emerson temperature discovery function that cannot be parsed as an integer, causing this ValueError:
Checkmk_Crash_kale_lab_54b25f28-30d6-11f0-bf2f-005056a12db3_2025-05-14_11-46-04.tar.gz

This results in a perpetural state of "unknown" for the discovery service, and an accumulation of crash dumps:

Proposed changes

There may be a better way to fix this, but it's difficult for me to know if e.g. changing the match string would break the original piece of hardware this check was written for. Python doesn't seem to have a quick or elegant way to comprehensively see if a string can be parsed as an integer other than to try and catch it if it fails, so that's what I opted for, here.

Thank you for your consideration!

[mo-ki]

Hi!
Thank you for your contribution!
The problem with this approach is that it masks all other problems as well. Is there any change to deal with a value like IB2 Temp 1 properly?
If that can't be done, I suggest to not detect the affected devices at all, by setting

detect=all_of(
    startswith(".1.3.6.1.4.1.6302.2.1.1.1.0", "Emerson Network Power"),
    not_ ...   # some condition to exclude affected devices
)

Any chance you can come up with a good condition on which the affected devices are?

[indigoparadox]

I agree that there are some pitfalls, so here's an alternate approach:

All of our devices are exhibiting the issue outlined above because they seem to be providing OIDs beyond those the check was originally designed for. These OIDs (1.3.6.1.4.1.6302.2.1.2.7.3.*) do not appear in any MIB I presently have access to (e.g. https://mibbrowser.online/mibdb_search.php?mib=EES-POWER-MIB), and looking closely at their values, it's difficult to tell if they're even providing valid data for temperatures above "Temperature 1".

With this being the case, I've rewritten the OID selection to explicitly confine it to 1.3.6.1.4.1.6302.2.1.2.7.1 and 1.3.6.1.4.1.6302.2.1.2.7.2, thereby limiting it to the (known good and valid) values the check was originally designed for and eliminating the crash when attempting to parse invalid data (because these OIDs return valid temperatures in a sane range on our devices).

What do you think?

Thanks!

[mo-ki]

There are many ways of doing this, obviously, but this way a test is failing that ensures we can deal with an emtpy input. I suggest to leave line 52 alone, add a comment, and do this:

def parse_...
    # we only use at most the first two lines because....
    return string_table[:2]

[mo-ki]

I like this approach better.
I just realize that this plugin is not using the system object ID nor the system description in its detect spec. ~35 of our ~1000 plugins do that. It's unfortunate, because it means that one adiitional OID is fetched during every discovery of any of your hosts. Would you share your system description with us? Maybe this is an opportunity to improve this :-)

[indigoparadox]

Certainly, the approach you outline in your review does seem to work well on our test device, so let's go with that.

The sysOID is: .1.3.6.1.4.1.6302.2.1

I hesitated to change any of the detection logic in the (probably naive) assumption that the original author was using a weird method of detection for a reason and I didn't want to break anyone's existing hardware that I may not have access to (especially if it's already behaving differently from the hardware I do have access to). If changing that does break anything, though, hopefully someone would submit an appropriate fix with a comment to this effect. 😌

[logan-connolly]

@indigoparadox thank you for the contribution!

You can find more details regarding when this patch will be released in Werk #18282.