Re: [Nagios-devel] Re: FreeBSD thread issues
Posted: Fri Aug 19, 2005 11:45 am
Christophe Yayon wrote:
> Andreas,
>
> in others words, it appears that Nagios is right and FreeBSD do not accept
> the 'fork method' choice of Nagios...
Most likely, yes. It does, however, appear that the spec should either
be fixed so that it is absolutely certain how a fork()'ing application
should behave, and they should definitely leave suggestions out of it,
since those have a way of being misinterpreted as recommendations.
Perhaps the threads library shouldn't bother with mutexes and just leave
it to the application coder to make sure a fork()'ed thread never runs
into a deadlock. That seems by far the most sensible (and for all
parties easiest) approach.
> The solution for our problem seems to have a specific Nagios patch for
> FreeBSD...
> Perhaps Nagios committer could do something, i will contact him ...
>
Maybe. It's entirely possible that it will be easier to migrate to a
one-thread-per-check solution rather than the current approach. It will
certainly shave a fair few cycles off the running of checks.
--
Andreas Ericsson [email protected]
OP5 AB www.op5.se
Lead Developer
This post was automatically imported from historical nagios-devel mailing list archives
Original poster: [email protected]
> Andreas,
>
> in others words, it appears that Nagios is right and FreeBSD do not accept
> the 'fork method' choice of Nagios...
Most likely, yes. It does, however, appear that the spec should either
be fixed so that it is absolutely certain how a fork()'ing application
should behave, and they should definitely leave suggestions out of it,
since those have a way of being misinterpreted as recommendations.
Perhaps the threads library shouldn't bother with mutexes and just leave
it to the application coder to make sure a fork()'ed thread never runs
into a deadlock. That seems by far the most sensible (and for all
parties easiest) approach.
> The solution for our problem seems to have a specific Nagios patch for
> FreeBSD...
> Perhaps Nagios committer could do something, i will contact him ...
>
Maybe. It's entirely possible that it will be easier to migrate to a
one-thread-per-check solution rather than the current approach. It will
certainly shave a fair few cycles off the running of checks.
--
Andreas Ericsson [email protected]
OP5 AB www.op5.se
Lead Developer
This post was automatically imported from historical nagios-devel mailing list archives
Original poster: [email protected]