On Tuesday, May 17th from 10:00am to 12:00pm the Nebula web interface and API interfaces will be unavailable for maintenance.
We will work on:
- Software and security updates
- Test live migration using new GlusterFS backend
- Configure compute nodes to use locally-mounted GlusterFS filesystems for glance images instead of downloading from controller (speeding up instance creation)
- Start bringing new storage online
- Work on bug with rabbitmq log rotation
Instances and networking to those instances will continue to work as expected during the maintenance time. Functions that try to provision new instances or volumes will not work, however.
Updated packages on network node
Tested live migration between two compute nodes and was able to evacuate a host, update its kernel, reboot, and then move everything back to that host. Ran into a problem with a directio lock file, need to test more.
Tested locally-mounted glance images and it worked great, although snapshots were failing. Since it's unclear if snapshots were failing because of the live migration or the glance change this change was undone as well.
These will continue to be tested in a special new availability zone.