Note that this merge may break things. The change from OpenDeployment
is subject to further review at a later time. The change introduces
openstack specific logic in an otherwise independent openssh recipe.
Change-Id: I568e4d2329a1d0e2670e1b0340afca4dddf4f23a
Define role related log attributes in role configs so that
recipe does not need to check its own role every time when
writing a role related config from template, because those
info would already have been in role definition and passed
down to node attributes.
Process monitoring can use the same mechanism.
Change-Id: Ifb10da356baf99d2d7e21c8d4d3d8df38fc81f58
There were also occurances that httpchk experienced timeout. Right
now we are using default value 2000ms. Try using 10s suggested by
the example config in openstack doc.
Related-bug: #1344144
Change-Id: I7a72d7a22a1845d1e2d480eee88246f980f87e6c
When backend gets extremely busy especially such complex system as
openstack, increase timeout to allow for longer tcp connnection time
before the backend is deemed dead.
Refer to: http://docs.openstack.org/high-availability-guide/content/
ha-aa-haproxy.html
Change-Id: I03f5937424dc77a94eec0c03a890668a15358e01
Related-bug: #1344144
ubuntu has an old apt cache upon a fresh installation, so some
package may not be in the cache yet because I noticed some dstat
installation failures due to no condidate found, so run apt-get update
first before installing packages.
Change-Id: I02cc51c45ab65a8eb9c915a87fad28aa4d844acb
https://bugs.launchpad.net/compass/+bug/1314742
Create network failed sometimes due to rpc timeout,
so need to know where the delay is.
Change-Id: I135d1153c6cf18e4fd2047e5f3e8c7ec61165936
Also increase the timeouts to see if any more Gateway timeout
failure in tempest.
Fix a bug in hosts snipet and also make log from haproxy tagged
with same hostname used by anamon.
Change-Id: I02ca18bb5ce0cdcd8a496672cb1bb5af179745be
Before running rsyslog restart make sure openstack services
start normally, otherwise, rsyslog restart will fail without
monitored dirs created by openstack services.
Change-Id: Icb3f91e97d8403130afa390fd4508a954cb41240
If the monitored log dir does not exist, rsyslog will abort,
so make sure only monitor the logs that are existent.
Change-Id: Ic0f265a6a85d1e50d1e6e044d18ccd94f6dd5937
