FAQ
Which ports are required to be opened on the VA/SN/LC servers to allow orchestrator to connect for upgrade?
This is documented below, port 3142 is required for apt-cache reachability which is caching proxy server for ubuntu packages, and 9183 is required for REST API calls towards the Director
What is the recommended order of upgrade?
The order should ideally be to upgrade all the cluster nodes first (VA nodes followed by Search nodes) one at a time, upgrade a node and move to the other only when the upgrade is successful, and once they've all been upgraded successfully you can upgrade the log-collectors one by one.
You can set the max-connections to 0 (in the config mode, and commit) on the log-collector which you are upgrading and set it back to original value post the upgrade - this way there are no connections on the log-collector during its upgrade.
You don't need to shutdown the van-driver on the log-collectors during the upgrade of the cluster nodes, since you would be upgrading one VA/VS node at a time there will always be redundancy available to cater to the incoming logs (if there are at-least 2 VA and 2 VS nodes in the cluster)
We are planning to upgrade using two or more orchestrator setup. In case of multi orchestrator environment, can we run three parallel upgrades on three VA Log's simultaneously triggered from three orchestrator?
It’s recommended to upgrade the VA/VS/Log-collectors one at a time to ensure the availability of redundancy to avoid data-loss, as explained in the answer above to the query about the order of upgrades
Will stopping the Driver on the LC’s and disabling auto start “ETL_AUTOSTART” in the vandriver.conf throughout the upgrade duration be enough to ensure avoiding data loss?
This is not required, please refer to the above question about the order of upgrade
What are the backups that we can take before the upgrade in case a rollback is required?
There is no provision to “rollback” an OS upgrade. If possible you can take a snapshot/backup of the VMs (if it’s a VM environment) involved. You can look to take any analytics/search “reports” that you deem are critical.
You can take a backup of the below files from each of the cluster/LC nodes
/opt/versa/scripts/van-scripts/vansetup.conf
/opt/versa/scripts/van-scripts/vandriver.conf
/etc/hosts
You can also take a config back as below
cli
show configuration | save backup.cfg
Also take a snapshot of “route -n” from the shell, as a reference for the route tables present on the system