Release 2.6.0 (code number 4290) includes several new features, an architectural improvement, and keep current updates.
Release 2.6.0 will be available for deployment to new servers from February 13th 2019. The Support Team will contact customers in order to schedule the date of server upgrades.
Architecture Change – ZDISKDD Removal
The Zynstra system architecture has been optimized and simplified by the removal of ZDISKDD.
This brings benefits to users as well as accelerating Zynstra product development.
For users and IT administrators, system start-up times have been improved thanks to fewer services being involved in disk and storage management. These functions move from ZDISKDD to ZHAPCON, reducing both the quantity of services starting up and the dependencies on services being ready before Custom VMs are able to boot.
Simplification of the architecture will improve all aspects of future product development that interact with server storage. In particular, server validation processes and on-going stability will benefit from this update.
Limiting Patch Download Bandwidth
For customers with limited WAN bandwidth, or who are particularly sensitive to saturation of the LAN or WAN, a rate-limit is now available.
By default, Zynstra servers will not limit the bandwidth used for download of Zynstra software patches. If required by the customer, a limit can now be configured on a per-server basis.
Custom VM Start-up Dependency
If you have applications running in multiple Custom VMs that have start-up inter-dependencies, a start-up sequence can now be implemented.
This configuration option can be requested from the Zynstra Support Team for set-up during server provisioning.
Custom VM Monitoring Agent Check
Monitoring agents installed inside Custom VMs provide two important features: Reporting the state of the VM and responding to graceful shut-down requests.
These agents may be deliberately or accidentally omitted from the VM, particularly where customers are deploying their own golden images. This can cause issues when a graceful shutdown is attempted, resulting in an unexpected hard shut down after a timeout period.
A new monitoring check now allows IT admins to check whether the agent is present in the Custom VM.
Configurable Retention Period per-VM
The option to set customized cloud backup retention periods for Custom VM data can now be set on a per-VM basis, rather than a single setting for all VMs on the server.
Configurable retention periods are an option for enterprise and ROBO customers hosting their own cloud storage as a backup target. Contact the Zynstra Support Team to discuss use of this feature.
Backup of Non-standard Custom VM Drives
Imported VM images, such as those created by P2V or V2V processes, may contain disk volumes in addition to the standard C: and D: drives used by all Custom VMs.
These additional drives can now be included in the Cloud Backup process.
Slow VM Death Fail-over
High availability failover is more reliable, for more VM failure scenarios.
This particularly relates to scenarios where the VM is observed to be failing over time, or failing slowly, rather than an immediate ‘death’.
Previously, if the kernel on one node reported that the system is degraded, but the other node can still detect its presence, the system would not failover. This could lead to the service being affected for some time and subsequently followed by a partial and unclean failover. With this change, degraded nodes will immediately reboot allowing a timely and clean failover
Critical Kernel Messages Saved in Pstore
Critical kernel messages such as panics can now to be saved in pstore. Critical logs on server reboot are retained for analysis.
This can provide insights into previous failures even when VSP logging is not enabled or has filled up to the point that interesting logs have been lost.
This release includes:
- Kernel 4.4.0-139
- Xen 4.8.4
- ZFS 0.7.12
- HPE ML110 Gen10 firmware