Software Release v1910
This is our second release for Ceph Nautilus, featuring a lot of smaller improvements and user experience enhancements.
Please read the upgrade notes at the end of this post carefully before upgrading.
New Server Image
There's a new server image with the latest Ceph 14.2.4 release. Like our release, this Ceph update brings a lot of small improvements and improved performance.
Changes
- Managed /etc/hosts in Docker so you always use the host name when working on a shell
- Managed ssh known hosts file for improved security when using SSH on the shell
- Better handling of large S3 buckets (more than 100 million objects)
- Manage abandoned S3 multipart uploads
- More reliable cluster imports by gathering more information about the existing cluster
- Improvements for multi-site S3 setups
- Support for importing Luminous and Mimic clusters
- iSCSI images are now resizable from the the iSCSI service UI
- Initial support for ceph-mgr modules: Ceph dashboard and balancer can now be enabled from the UI
- Improved templates: edit kernel command lines in a separate template, more helper functions
- Detect more common configuration errors: bad timezone or date settings on a server, bad disk configurations, bad pool configurations, and more
- Add custom notes to servers and disks (e.g., mark a disk as due for replacement)
- More reliable OSD detection from old clusters (ceph-disk)
- Graceful handling and detection when some network cards are changed
- Better IPv6 handling
- The number of PGs in a pool can now be reduced
- Lots of small improvements and bug fixes
Upgrade Notes
As always: ensure that you have a working backup (like our encrypted cloud backup) before upgrading the container.
When upgrading from 19.01 or earlier: There's one additional step necessary when upgrading the container due to the upgrade to Debian Buster: please ensure that the MySQL database is shutdown correctly. The first command accomplishes that:
docker exec croit service mysql stop # only necessary when running 19.01 or earlier
docker rm -f croit
docker run --net=host --restart=always --volumes-from croit-data --name croit -d croit/croit:1910
Upgrading without running the first step might leave you with a container that fails to start the database. If you run into this: simply revert to the old container by repeating the commands with the old version
S3 Users
S3 users must upgrade to Ceph Nautilus after upgrading to croit 19.10 due to internal incompatibilities in Ceph, not upgrading Ceph after upgrading croit will disable some functions in our dashboard. S3 applications will continue normally.
ISCSI Users
The new Nautilus images come with Ceph iSCSI 3.x. Please reboot all iSCSI servers with the new image after upgrading to Ceph Nautilus. Some new features in iSCSI are only available after doing so, running mixed images will affect failover capabilities.
NFS Users
We still recommend to use NFS version 3 instead of version 4 when using NFS due to reliability issues encountered during testing.
We recommend all new setups to use iSCSI or SMB instead of NFS if possible.