SNMP: Improve parsing of OIDs containing IP addresses

[rousskov]

Broken since 2007 commit cc192b5 made a false "4 or 16" assumption,
possibly influenced by bad assumptions in earlier client_Inst() code.

Also improved const-correctness to get fixed oid2addr() code to compile.

[rousskov]

It would be better to use i6addr variable instead of guessing its type inside sizeof(), but that would break symmetry with the above IPv4 code unless we introduce more out-of-scope polishing changes.

I also duplicated odd if line formatting to keep IPv4 and IPv6 code as similar as possible without introducing even more changes.

[rousskov]

@eduard-bagdasaryan, could you please check Anubis configuration? AFAICT, it was not updated when FreeBSD tests were removed in 8c76249, and Anubis is now waiting for the staging tests that will never run.

[kinkie]

@eduard-bagdasaryan, could you please check Anubis configuration? AFAICT, it was not updated when FreeBSD tests were removed in 8c76249, and Anubis is now waiting for the staging tests that will never run.

I've updated the config.
I have a small feature request for Anubis: in place of (or in addition to) printing the full list of successful checks, print their count - the log contains the list of requirements but too much detail the results, it's hard to interpret it

[rousskov]

@eduard-bagdasaryan, could you please check Anubis configuration? AFAICT, it was not updated when FreeBSD tests were removed in 8c76249, and Anubis is now waiting for the staging tests that will never run.

I've updated the config.

AFAICT, it did not help:

PR2419 (head: 6c03eb staged: 8f73b1) ... did not merge: waiting for staging tests completion
.... "time":"2026-05-20T21:03:43.202Z"

I have a small feature request for Anubis: in place of (or in addition to) printing the full list of successful checks, print their count - the log contains the list of requirements but too much detail the results, it's hard to interpret it

This request is best submitted at https://github.com/measurement-factory/anubis/issues but is it possible that this is already implemented? I see the following in the log that seems to contain the counters you are asking for:

PR2419 ... staging status details: ... Staging expected/required/optional: 101/95/0

The "expected" 101 count matches the updated staging_checks setting in Anubis configuration for Squid. GitHub shows 94 staging checks (all are considered to be required IIRC). Anubis copies PR Approval status check, bringing that count to the reported "95". Anubis then waits for the missing 6 checks...

That is just my quick interpretation -- I did not check any details. @eduard-bagdasaryan, please correct me if I got it wrong.

[kinkie]

@eduard-bagdasaryan, could you please check Anubis configuration? AFAICT, it was not updated when FreeBSD tests were removed in 8c76249, and Anubis is now waiting for the staging tests that will never run.

I've updated the config.

AFAICT, it did not help:

PR2419 (head: 6c03eb staged: 8f73b1) ... did not merge: waiting for staging tests completion
.... "time":"2026-05-20T21:03:43.202Z"

I have a small feature request for Anubis: in place of (or in addition to) printing the full list of successful checks, print their count - the log contains the list of requirements but too much detail the results, it's hard to interpret it

This request is best submitted at https://github.com/measurement-factory/anubis/issues but is it possible that this is already implemented? I see the following in the log that seems to contain the counters you are asking for:

PR2419 ... staging status details: ... Staging expected/required/optional: 101/95/0

The "expected" 101 count matches the updated staging_checks setting in Anubis configuration for Squid. GitHub shows 94 staging checks (all are considered to be required IIRC). Anubis copies PR Approval status check, bringing that count to the reported "95". Anubis then waits for the missing 6 checks...

That is just my quick interpretation -- I did not check any details. @eduard-bagdasaryan, please correct me if I got it wrong.

You seem to be right.
Currently the staging test has 95 tests. This means that it was not only the freebsd test that has been removed.
Okay, I'll tune anubis to pass with 95 successful results on staging. We are due to update tests soon anyway, as both Ubuntu and Fedora have released new versions.

[rousskov]

Okay, I'll tune anubis to pass with 95 successful results on staging.

PR2419 (head: 6c03eb staged: 8f73b1 history: load,merge,update) staging checks succeeded

After your adjustments, Anubis went further, but it ran into other problems. At this time, I think it is best to give Eduard a chance to check things out.