Place update_state methods out of db session;
Match package installer name by regex.
patch2: fix style
Change-Id: I26d3f21e2d3c022ebe2b4635583c33074434ae46
temporarily commented out run_in_session in util
Separated DB tables for clusterhost and host state.
Change-Id: Ib4e4c6c1931bbbf62b0b77bc994bffecb0fe27bd
Create virtualenv for compass, point compass site packages to virtualenv,
make virtualenv accessable for apache.
Change compass executable scripts to use python in env to be virtualenv
compatible.
Make progress_update daemon use virtualenv.
Comment out some health checks that use packages not available to pip, eg,
yum, apt since we have other ways to check packages any way.
Conflicts:
conf/compassd
install/compass.sh
install/dependency.sh
install/prepare.sh
misc/apache/compass.wsgi
requirements.txt
Change-Id: Ib48633601bf238ffc660c4694834897edbe45e07
now start from *.*.*.10 but test slaves where compass server
is can be out of 10 so a conflict might happen, so increase
to 50.
pin pycrypto version to 2.0.1 and below
Keystone client grizzly version only work with pycrypto below 2.0.1,
so pin pycrypto to 2.0.1 and below.
Closes-Bug: 1321032
Closes-Bug: 1321056
Change-Id: I2f72937ee128d34ce2eb02f35edff700eab686a5
To process logs by os_loganalyze on log server, the format
need to be the same as in original log files.
Change-Id: I07935eae0d29005725a27f8b4b4056219ccf7005
Float ip needs route to private network so floating ip can
be associated to a vm.
Also make rsyslog config apply for logs from haproxy.
Change-Id: I628bc8d6c7486301e1a9aba2cd78080285e33d8d
Sometimes when running volume related tempest tests, there is
change that such error occurs: "Failed to create ALL necessary
volumes to run test". It might because a race in deleting and
creating volumes. Anyway, 5G cinder volume is enough but not by
a margin. Give it a little more space.
Change-Id: Id129c21461fd027e4eddf9ea7aee24f74154dee7
If nova-compute and nova-conductor are on different nodes
and n-compute is up before n-conductor, n-compute will retry
to register health but there will be increasing wait time
between retries, so make sure n-conduct sees available compute
nodes before starting tempest.
also added a parent http proxy to speed up installation and
a dependency for tempest.
Change-Id: Ica46c66bc8b7589915490555d86674eb49c0db1c
also add a skip-broken flag to yum update for the time-being
because a syslinux dependency: syslinux-nonlinux was not found
in centos mirrors. It is on the way though.
https://www.centos.org/forums/viewtopic.php?f=14&t=45933
Need 'yum clean all' to update metadata and get the latest package
list from mirros.
Fix a bug in prepare.sh. it didn't fetch the gerrit refs.
Restore some package pre-load scripts in prepare.sh to make CI faster.
Change-Id: I66f06bdfb801204fb956c7a264c98de879322744
