Hi Tomas,
The patch looks good to me.
Just a minor remark.
ldap_inst->exiting=TRUE and signaling the watcher thread is the same
action. Ideally the signal handler would set 'existing=TRUE', but there
is no nice way for the signal handler to retrieve/set the 'existing'
flag. Do you think we could move 'ldap_inst->exiting=TRUE' and
pthread_kill in a same wrapper function (for example watcher_shutdown).
thanks
thierry
On 12/19/2016 01:04 PM, Tomas Krizek wrote:
Hi Thierry,
could you please take a look at this bind-dyndb-ldap patch? I was trying
to fix https://fedorahosted.org/bind-dyndb-ldap/ticket/149
I wasn't able to reproduce the issue, but I think the problem is fixed
now. Petr Spacek was helping me with this, but I think it would be good
if you could take a look as well.
We were able to identify two causes: a) isc_thread_create fails and
ldap_inst->watcher is undefined (fixed by setting it to 0) and b) the
thread terminates for some reason (i.e. some checks fail) and then a
signal can't be sent to it. This was addressed by removing the REQUIRE
and logging an error instead.
Thanks.
--
Manage your subscription for the Freeipa-devel mailing list:
https://www.redhat.com/mailman/listinfo/freeipa-devel
Contribute to FreeIPA: http://www.freeipa.org/page/Contribute/Code
