The current template assumes that websockets are at the base of a URL
but that is not true for our buildmaster. This patch adds remotepath
to the end of the websocket url if remotepath is defined.
Currently update-fullfiletimelist is called from 4 machines:
* rawhide-composer: after rawhide compose
* branched-composer: after branched compose
* secondary01: in a cron every hour updating only /pub/alt as it can change anytime.
* bodhi-backend01: after ever updates sync that has content synced
We need bodhi-backend01 to use the same lock as rawhide/branched composers since they
update the same modules and we don't want them to step on each other.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
This was offered by Alessandro Lorenzi <alorenzi@alorenzi.eu> as a fix
to deal with our inability to monitor the certgetter after
reboots. Thank you very much for this work.
Signed-off-by: Stephen Smoogen <smooge@redhat.com>
If a YAML dictionary is used, this gets turned into a string before
being handed to the rabbitmq_parameter module. The problem is it uses
single quotes instead of double quotes, so when the module attempts to
JSON-deserialize it, it explodes.
Add in drives45 to hand over to Kevin for 45drives evaluation.
Remove ipv6 from inventory so system can be cleaned out. This box should never have been on the frozen category.
No idea why that file got renamed.
In recent months/years, pungi has started to use more and more memory when gathering package sets.
So, give it enough memory to avoid the OOM killer killing composes.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
The various openshift tools get hit by various crawlers and do not send
a robots.txt. This seems to be due to the balancer code used to send
back to the nodes. This forces the proxy robots.txt to be honored
always.
Some of fedbages query to datanommer are really slow, that makes
the consumer enable to catch up with all the messages on the bus
Giving it 2 more threads improves the situation.
Signed-off-by: Clement Verna <cverna@tutanota.com>
When adding a host to the cloud, it needs to be added to the inventory
file. The data in the file was organized by alphabet which meant that it
wasn't clear that hardware was also in this group. Reorganized the file.
Added data to the virthost data to make it clearer that it is like the
other cloud hosts. Removed it from the vpn mail group because it is not
on the vpn. We will need to create a cloud postfix group for these boxes.
fewer modules in Fedora. Our Koji has capacity to run many more builds.
Currently we have 158 Koji builders in default channel that are able to
run RPM builds. Their total capacity is 566. Therefore it should be safe
to increase MBS NUM_CONCURRENT_BUILDS to at least 100. Especially since
Koji builds submitted by MBS have lower priority than default for
non-modular component builds submitted by packagers, so
packager-submitted builds will tak precedence over builds submitted by
MBS.
Signed-off-by: Igor Gnatenko <ignatenkobrain@fedoraproject.org>
These aarch64 virthosts will have copr builders on them, so
we need to setup a copr user and add their ssh keyfor access,
then make sure they are in the libvirt group so they can manage vms.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
