Change Id8b34407fcc5b1705f3db17a5fc26a4a79034baa added the original
incorrect .asc key here which is for key "2C52609D Docker Release
Tool". However it also setup the various puppet bits with the correct
"0EBFCD88 Docker Release (CE deb)" key. It is unclear how this was
working, if at all.
Anyway, this is the correct 0x0EBFCD88 key
Change-Id: Ie21ebb8714d73461afda2ccfc57eb66babf931ce
The upstream mirror stopped using that old address a while back. The new
mirror has the latest packages. We are currently 5 months out of date.
Change-Id: I2b938da2cb89c996d3e9d8173377f44d2a31f710
This reverts commit efb3e4c6a8a681e09ad41273da9ce1c1e5d4faf2.
Change-Id: I57b117fab45603a9ba7213b43b6efbdc0c453509
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
Set ZooKeeper's autopurge.purgeInterval to 6 hours for
nodepool.openstack.org to limit the number of ZK snapshots kept.
These will quickly fill the disk in production if this is not
enabled.
The purge_interval class parameter determines the frequency in hours
to look for and purge old snapshots, defaulting to 0 (disabled). The
number of retained snapshots can be separately controlled through
snap_retain_count but already defaults to a sane value of 3 (the
minimum supported for autopurge.snapRetainCount).
Change-Id: If84094641add539a815718ae604729fe6f0545cd
We are having load issues on our current 8 zuul-executors, bring
online 2 more to help spread the load. It is possibl we need to do
this a few times until we find the proper amount.
Change-Id: I84b366e6c89ff3b7c4dfd5a363047b8573f4dd95
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
We want to start collecting statsd metrics again.
Change-Id: Ida577643e9c05ded1e953de7e2e9ed8f314d8981
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
With the new upgrade of Gerrit, the class name has changed
for the search bar resulting in a longer-than-expected bar.
This patch updates and points it to the right class so that
we restore the old search bar size.
Change-Id: I4eee34198c3d46954d5e0677324c3338320dcdee
We only have 100GB disk for tripleo-test-cloud-rh1, between AFS and
apache cache, we've filled the disk. Drop down to 40GB until
tripleo-test-cloud-rh1 is removed, then we can revert.
Change-Id: I6e3d1b6f3202b375853919d1920bd945fe035855
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
Cloud-init is changing our hostname on servers when we reboot. Stop
this from happening by disabling it.
Change-Id: Ia825a7823d7099870885636e0adb4134c5568715
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
Apparently we might need to keep all account entries in memory.
We currently have 26k accounts, so this gives us head room.
Change-Id: I33280ecc4c0bb24600ca78814e0a953d9dc696d7
We accepted these parameters at a system-config level but were not
passing through to puppet-gerrit manifest which actually writes out the
template file. Add these to the ::gerrit class instantiation.
Change-Id: I2abd24727ba39e3743b2d6b6ec279c70faf24d10
We're currently at 1024 entries with a 64% hit rate. Increase to
2048 to hopefully speed things up.
Change-Id: Ifeaf282a7cf0bd91342cc4ec24117ef64ed1054f
Depends-On: Iedb9fae62b133d76efeac5c8ecffde49704828c2
It was noticed that review.o.o doesn't have the "(cgit)" links next to
commit/parent hashes in the commit box.
It looks like the inbuilt gitweb integration is returning 404 for some
reason after the upgrade. One option is to use the cgit integration
instead. This seems to make some sense as cgit is behind a
load-balancer and may perform better than local cgi calls?
Note review-dev uses cgit like this, but that may be just
unintentional.
Change-Id: I86d8e724b22ee349f58540864888926d766daf1d
It ran up against the 30g limit within a couple of hours after
the 2.13 upgrade. Give it more headroom to hopefully let it
function longer.
Change-Id: I41fb5929c99b82858f7dac28d89497db332bdbd8
This change can't quite do the upgrade itself unfortunately. Instead the
upgrade should be performed as described at
https://etherpad.openstack.org/p/gerrit-2.13.-upgrade-steps then this
change should be merged before puppet is reenabled on that node. This
ensures that puppet wants to see the post upgrade state preventing it
from undoing anything.
Change-Id: Ide9e89542ae4203d70b5480475993d79259a3840
We don't actually use drpm in centos-7 nodes, so stop rsyncing them.
Change-Id: I70d198bd7d532ce2f9c4607059cb2a795025dd91
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
We incorrectly created known_hosts on zuulv3 mergers.
Change-Id: I1e0739679c202d024edb056fa9874849856f03b5
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
Bring online 4 more zuul-executors for zuulv3.o.o.
Change-Id: Ice42333f5c06f3ed0d32eeda65d09794a523f355
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
We're now launching xenial for control servers, lets update the
defaults.
Change-Id: I14dc26673c290ae37b7a9ef016d7a343d2763efe
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
Because this is shared with zm01-zm04, we only got the jenkins user
SSH keys. So we need to also add zuul user ssh keys, but we can
refactor this once zuulv2.5 is done.
Change-Id: I9c9adb1943b165014de516849de79d999419069d
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
Rackspace is now providing a partition table on /dev/xvde1, so check
to see if a filesystem exists, and if now, delete it and rebuild.
Import our logic from devstack-gate role to fix swap issues too.
Change-Id: I81571489cab121159f48a9437a24388696db331c
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
