Hi Grimur,
The issue you did hit was solved last December and will be part of
upcoming release 1.7.3:
https://github.com/pmacct/pmacct/commit/56c498e60043c868131d64404f3c3a8f338ea406
Until the release will be available, you could use 1.7.3-rc2 or master
GitHub code. It is currently left to you to implement a recovery
mechanism in case of crashes - consider spawning pmacct daemons off of
a systemd service would do plenty of the work for you in this sense,
ie. checking (sub-)processes liveliness and restarting the daemon if
negative.
Paolo
On Wed, Apr 10, 2019 at 09:28:50AM +, Grímur Daníelsson wrote:
> Hello everyone
>
> I had this error last night which caused the amqp plugin to crash.
>
> nfacctd: plugin_common.c:442: P_cache_insert: Assertion `!cache_ptr->stitch'
> failed.
> WARN ( default/core ): connection lost to 'ingress-amqp'; closing connection.
>
> After this error occurred the service kept on collecting as if nothing
> happened but it didn't export anything and so we
> lost all ingress traffic for the night.
>
> We have another amqp exporter that is for egress traffic which kept trucking
> along just fine.
>
> Is there no recovery mechanism in place if an exporter crashes or loses
> connection?
> Isn't it better to crash the program with a fatal error in these cases?
>
> Thanks for the help.
>
> Regards, Grimur
> ___
> pmacct-discussion mailing list
> http://www.pmacct.net/#mailinglists
___
pmacct-discussion mailing list
http://www.pmacct.net/#mailinglists
