unsure if both the Init SSL messages should go to upsdebugx (or neither, really).

One alternative is to use debug level 0 so it always pops up on stderr like now, but does not hit the syslog.

Asked community on mailing lists - if insecurity reminders were deemed useful.

Question: which logs does it clutter - syslog (thanks to upslogx used) or some unhandled stderr in munin?

munin writes unhandled stderr to syslog (*/5):

Sep 16 09:45:11 aquos munin-node[2781871]: 2022/09/16-09:45:11 [2781871] Error output from nut_misc:
Sep 16 09:45:11 aquos munin-node[2781871]: 2022/09/16-09:45:11 [2781871] #011Init SSL without certificate database
Sep 16 09:45:12 aquos munin-node[2781871]: 2022/09/16-09:45:12 [2781871] Error output from nut_volts:
Sep 16 09:45:12 aquos munin-node[2781871]: 2022/09/16-09:45:12 [2781871] #011Init SSL without certificate database

i suppose the munin plugin could be modified to redirect stderr to /dev/null, but that'd also remove useful errors. my reasoning was that adding a grep -v pipe would add extra overhead. i think it'd also require redirecting stderr to stdout which would mess with the parser if other errors were emitted.

thanks for putting a feeler out.

edit: it does not write to syslog by default. i think it writes to /var/log/munin/munin-node.log.

FWIW, a link to stirred-up discussion : https://alioth-lists.debian.net/pipermail/nut-upsdev/2022-September/007761.html

So far it seems the proposition is reasonable, but this quiescing should be non-default. Might be activated by an envvar and/or command-line option. since upsc (or for that matter other clients -- consumers of this library -- like upsrw and upscmd) do not have any config files.

i think an env variable might work. adding a new cli flag wouldn't really be usable if it gave a syntax error when it's unsupported.

i did look pretty closely at how NSS is initialized, and the conclusion i came to is that simply changing the verbosity of the message was the most straightforward thing to do.

Thanks for raising the concern about "a syntax error when it's unsupported", so I posted a PR for an envvar-based solution. Can you check if that looks good for your purposes?

yep, it should work. left a note there as well. thanks!