From this version, UVS supports updated UVS software sub-versions. Customers can upgrade new features or bugs fixed when they are generally available without waiting for the full version UVS release. In this release, we also update the Ceph version to Nautilus v14.2.9 to improve the stability.

Major Changes in v2.14.17

Upgrade Ceph from Nautilus 14.2.5 to 14.2.9. Some critical issues are fixed and listed as follows. Please refer to the Ceph Nautilus Release Notes.

• Fix the ceph-mgr heartbeat issue
• Fix a couple of security issues.
• Improve the ceph-osd usage of memory.
• Improve the stability of Cephfs.

UVS manager new features

RBD Mirroring

• Add installation wizard.
• Add/Del rbd-mirroring daemons.
• Add/Del RBD Mirroring peers.
• Support RBD Mirroring promotion & demotion.

Deployment

• Users can change the timezone of the UVS Manager and push the timezone setting to all nodes.
• Support subversion firmware update of UVS.
• Cleanup OSD disk and swap while removing OSD.

RADOS Gateway

• Enable Object Gateway function in the Ceph-MGR dashboard when the user creates an RGW service.

Ceph Cluster

• Restrict the user to create a pool with a PG number power of 2.
• Restrict the user to select the “rbd application” pool while creating RBD images.

Monitoring

• UVS Prometheus and node, ceph exporter service.
• Support the Cpeh-MGR Zabbix module.
• UVS now collects output logs of the operations.

Other Enhancements

• UVS mounts the M.2 SSD disk on the path /m2, and put more data into the SSD instead of the eMMC flash.
• Show the Ceph-MGR status on the Dashboard and Node pages.
• Add the “restart” button for the Ceph-MGR daemons for the case Ceph-MGR service stuck.
• Use ETCD to sync the notification counts.
• Limit resources to Ceph service with systemd.

 Bugs Fixed

• Firmware update function should avoid installing .upd files design for another distro.
• Fail to move Ceph monitor data to M.2 SSD.
• Tune the TCP behavior for improvement of Ceph slow requests & TCP read block issue.
• Failed to activate ceph-mon after removing an unhealthy mon node.
• Mars400 network has slow performance and holds lots of time_wait at monitor node.
• Check the SSD alignment.
• The function “Push RPM” can’t push itself.
• Fix updateetcd service logs a large unused message in the journal.
• Fix sysusage, and cephusage services log unused error message.
• Remove unused RPM packages.
• Update firmware function fails when ms model time is inaccurate.
• Extend the timeout period when upgrade.
• Improve the compatibility of external RGW SSL.
• The CephX user's name didn't match the real user's.
• UVS Openstack feature didn't set the pool application.
• Fix updateetcd service error.

Upgrade Guide for the upgrade from Version 2.14.16 to 2.14.17

Upgrade the UVS is simple. You just push the uploaded update file to selected nodes. You can select multiple nodes for one push update. UVS will do the rolling update one node at a time. There will be no downtime during the upgrade. But, please do follow the following procedure. 

The steps to completely upgrade the software are

1. Upgrade the first monitor node by the firmware update feature of the UVS manager. The first monitor node is the node you are using UVS manager for the upgrade.
2. Upgrade the rest of the monitor 
3. Upgrade the OSD nodes.
4. Upgrade MDS nodes if you have MDS nodes in use.
5. Check the UVS software versions.
6. Upgrade the external RADOS gateways.
7. Upgrade the external iSCSI gateways.
8. Update the cluster timezone.
9. Finally, please upgrade your ceph clients.

We have a video about the upgrade procedure at the end of this article.