Due to a previous hardware issue (https://lists.centos.org/pipermail/ci-users/2020-October/002124.html), we had a workaround in place that is still in place but is actually blocking a reorg of the infra used behind ci.centos.org.
We are announcing a maintenance window that will permit us to move network switches, and reconfigure the network properly , allowing also us to redeploy/load balance our workload across remaining available infrastructure.
Migration is scheduled for """"Tuesday August 17th, 1:00 pm UTC time"""". You can convert to local time with $(date -d '2021-08-17 13:00 UTC')
The expected "downtime" is estimated to ~60 minutes , time needed to migrate services, reconfigure network switch and restart the nodes connected to new network switch.
Impacted services in CI env:
- duffy API (unavailable during the move to other hypervisor, so normally unavailable for ~10min) : no way to request a duffy node for bare-metal tests - jenkins (aka ci.centos.org) and all agents connected to it - cloud.cico (opennebula infra used to provide Virtual Machines for CI tenants)
Thanks for your comprehending and patience.
on behalf of the Infra team,
On 09/08/2021 15:36, Fabian Arrotin wrote:
Due to a previous hardware issue (https://lists.centos.org/pipermail/ci-users/2020-October/002124.html), we had a workaround in place that is still in place but is actually blocking a reorg of the infra used behind ci.centos.org.
We are announcing a maintenance window that will permit us to move network switches, and reconfigure the network properly , allowing also us to redeploy/load balance our workload across remaining available infrastructure.
Migration is scheduled for """"Tuesday August 17th, 1:00 pm UTC time"""". You can convert to local time with $(date -d '2021-08-17 13:00 UTC')
The expected "downtime" is estimated to ~60 minutes , time needed to migrate services, reconfigure network switch and restart the nodes connected to new network switch.
Impacted services in CI env:
- duffy API (unavailable during the move to other hypervisor, so
normally unavailable for ~10min) : no way to request a duffy node for bare-metal tests
- jenkins (aka ci.centos.org) and all agents connected to it
- cloud.cico (opennebula infra used to provide Virtual Machines for CI
tenants)
Thanks for your comprehending and patience.
on behalf of the Infra team,
Just to inform that announced/scheduled maintenance is now complete and all services moved (or restored)
If you encounter any issue (for example if something was done at your side in your workspace and that we weren't aware of), please create a ticket on https://pagure.io/centos-infra/issues
Kind Regards,
-
Fabian Arrotin