View source for Open-E: Jovian Deployment Notes

[[Open-E: Jovian|Open-E Jovian Landing Page]]

== General Notes ==
* Jovian does not currently support at rest encryption. Open-E are intending to add ZFS native encryption in the 1.0 release update 30 planned for late 2020.
* The configuration of these systems including clusters is complex. It is strongly recommended you build the customer environment in a virtual environment in advance to confirm validity of Jovian configuration regardless of experience.

== Deployment Validation ==
* Validation process [https://www.open-e.com/site_media/deployments/documents/Open-E_JovianDSS_Deployment_Acceptance_Tests_Guide_v1.6.zip]
** The validation process among other things includes internal and external performance tests.
** Fail-over tests ranging from managed planned fail-overs to a system power cut.

== Cluster Fail-over Testing ==
* A Jovian cluster is active/passive meaning the fail-over is not seamless taking approximately 10 seconds for the passive node to take control of the pools and another 20 seconds for the pools to become available to clients via iSCSI. In Windows disk management the iSCSI disks may disconnect but they will re-connect by default in approximately 210 seconds. This can manually be forced from within the iSCSI initiator interface by clicking 'refresh'. This timeout may also be editable in the Windows registry.
* When performing cluster fail-over tests which includes performing the 'move pool' operation you may receive the following error, this error will usually resolve itself within 30 seconds and occurs when clients are using the specified pool:
[[File:Pool-last-accessed-by-another-system.PNG]]


== Access ==
* By default no external access to the Jovian console is available (this can be configured). To access Jovian console interface access via IPMI remote console.

== Networking ==
* Inactive interfaces, denotes the interface has been disabled in Jovian.
* Each interface MUST be on its own subnet.
** 2 or more interfaces on the same subnet may work for a while but eventually access will be lost requiring disabling all but one interface via the console to regain access.
* Node-A and Node-B bonds must be configured as fail-over active/passive when nodes are added to a cluster otherwise the cluster nodes will fail to bond and you will receive the following error, requiring the deletion and re-creation of the bond interface with load-balancing type set to active/passive:
[[File:Node-binding-error-load-balance-type.PNG]]
* Ring/ping interfaces can be bonded as active/active.