Probe Failure Reasons in Blackbox Exporter

[Tobias Guggenmos]

Hi,

I'm a user of blackbox exporter who uses it to send Alerts about unreachable services (the common use case i guess).

One of the most substantial pain points for me is, that the results of failed Probes do not provide information about why probes failed in a way, that is easy to use in Alerts.

A workaround is to check the Blackbox Exporter Debug Logs, but having that information present in metrics would be much more convenient.

There has been an effort to work towards partially addressing this situation, but it appears to have stalled: https://github.com/prometheus/blackbox_exporter/issues/1077

I would like to revive the work on addressing this issue and am willing to contribute both code and design proposals.

Do you have any advice on how to get such stalled feature proposals going again?

Best,

Tobias

[Bjoern Rabenstein]

On 03.12.24 05:29, Tobias Guggenmos wrote:
>
>
> There has been an effort to work towards partially addressing this
> situation, but it appears to have stalled:
> https://github.com/prometheus/blackbox_exporter/issues/1077

Hmm, yeah, it looks the PR linked to this issue never got any feedback
from the blackbox_exporter maintaners. This sometimes
happens. Maintainers are just humans, and sometimes they drop their
activity level.

> I would like to revive the work on addressing this issue and am willing to
> contribute both code and design proposals.
>
> Do you have any advice on how to get such stalled feature proposals going
> again?

It depends why it stalled. Sometimes features are controversial, in
which case it is better to first get some consensus from the
maintainers or the wider community that the feature is wanted (and in
which way). For involved things, this might require a formal proposal
AKA design doc via https://github.com/prometheus/proposals .

However, in simpler cases, the person working on it initially just
didn't have the time to finish it, in which case you could just pick
up the work.

In other cases, the effort stalled because there was no feedback from
the maintainers (the current case smells like that, but there might be
context that we cannot see from the GH issue and PR). In this case, I
would try to get in touch with the maintainers and find out if it was
just an oversight or there were concerns from their side (ask them
directly on the GH issue, with @-mentions, and if that doesn't help,
using channels like this mailing list or the #prometheus-dev channel
on the CNCF Slack might help).
--
Björn Rabenstein
[PGP-ID] 0x851C3DA17D748D03
[email] bjo...@rabenste.in

[Tobias Guggenmos]

Thanks a lot for the tips, they seem to have helped.