Currently, rhel-osp-director configures the mysql proxy in haproxy.cfg to have client and server timeout set to 0. This is incorrect. Doing this will result in the following warning: [WARNING] 238/115010 (13878) : config : missing timeouts for proxy 'mysql'. | While not properly invalid, you will certainly encounter various problems | with such a configuration. To fix this, please ensure that all following | timeouts are set to a non-zero value: 'client', 'connect', 'server'. [WARNING] 238/115010 (13878) : config : missing timeouts for proxy 'redis'. | While not properly invalid, you will certainly encounter various problems | with such a configuration. To fix this, please ensure that all following | timeouts are set to a non-zero value: 'client', 'connect', 'server'. The mysql timeouts should be >60m. I am not sure that the appropriate timeouts for redis should be.
Patch sent upstream for MySQL: https://review.openstack.org/223223
Emilien, what about redis? Is that going to be fixed in this bug too? or should we split the bug?
The upstream patch for mysql (in puppet-tripleo) is merged, I am going to clone this to cover the redis half, so mysql can move forward
*** Bug 1281353 has been marked as a duplicate of this bug. ***
Turns out this was only merged on master, it has now been cherry-picked back to liberty
Ryan, I have 3 questions for you on this: 1. The Version is set to 7, but the target release is 8 - did you see this bug in 7 or 8? I am not sure where to mae sure any fixes land (7 seems less likely since the bug is just a warning versus an error, but if it gets enough priority, that could change of course) 2. I see there was a change in the haproxy defaults[1] that did not exist when you saw this, I wonder if this warning would still occur if retested with current osp8? Seems to me like that would at least satisfy the warning, though we may want to keep the patch I just set up for backport anyway for mysql 3. Where did these errors get output? And is there any config change needed on an osp install to make them show up on a test system in the same place (so someone can try to reproduce) [1] https://github.com/openstack/puppet-tripleo/blob/stable/liberty/manifests/loadbalancer.pp#L40
(In reply to Jason Guiditta from comment #9) > Ryan, I have 3 questions for you on this: > > 1. The Version is set to 7, but the target release is 8 - did you see this > bug in 7 or 8? I am not sure where to mae sure any fixes land (7 seems less > likely since the bug is just a warning versus an error, but if it gets > enough priority, that could change of course) I don't remember. If anything has a timeout set to 0, that is wrong and needs to be fixed. > 2. I see there was a change in the haproxy defaults[1] that did not exist > when you saw this, I wonder if this warning would still occur if retested > with current osp8? Seems to me like that would at least satisfy the warning, > though we may want to keep the patch I just set up for backport anyway for > mysql The config I looked at had timeouts explicitly set to 0 in the proxy definition. That will override all defaults, so no -- setting defaults will not make a difference. > 3. Where did these errors get output? And is there any config change needed > on an osp install to make them show up on a test system in the same place > (so someone can try to reproduce) It depends. If you're starting it from the command-line you should see them on stdout. Else you should see them in the system logs.
Using director 8.0 puddle - 2016-03-11.1 ,mysql have 90m for both client/server timeout, and there's no timeout set for redis: listen mysql bind 172.17.0.11:3306 transparent option tcpka option httpchk stick on dst stick-table type ip size 1000 timeout client 90m timeout server 90m listen redis bind 172.17.0.10:6379 transparent balance first option tcp-check tcp-check send info\ replication\r\n tcp-check expect string role:master [stack@puma33 ~]$ rpm -qa|grep puppet openstack-puppet-modules-7.0.13-1.el7ost.noarch
Please note the redis half of this has been split into a separate BZ (1311300) so if this output above is verified, please verify the related one as well.
(In reply to Asaf Hirshberg from comment #13) > Using director 8.0 puddle - 2016-03-11.1 ,mysql have 90m for both > client/server timeout, and there's no timeout set for redis: > > listen mysql > bind 172.17.0.11:3306 transparent > option tcpka > option httpchk > stick on dst > stick-table type ip size 1000 > timeout client 90m > timeout server 90m > > listen redis > bind 172.17.0.10:6379 transparent > balance first > option tcp-check > tcp-check send info\ replication\r\n > tcp-check expect string role:master > > > > > [stack@puma33 ~]$ rpm -qa|grep puppet > openstack-puppet-modules-7.0.13-1.el7ost.noarch The mysql timeouts are fine. Since redis does not specify timeouts, it will use whatever timeouts are set in the default section (not shown here).
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHEA-2016-0603.html