To fix a bug related to host reboots, starting with version 890 of the Replicated-based on-premises Sysdig platform, new TCP ports are exposed from backend containers to the host level:
These take the place of random ephemeral port numbers that were exposed in version 800 and earlier.
Before upgrading to version 890, ensure that no other services you run on your hosts are currently using these new default ports. If there is a conflict, read below for information on how to change the ports via the admin console.
In version 800 and earlier, random ephemeral port numbers (32768 and higher) were auto-assigned to expose the API/Collector backend containers to the host level. The load balancers for API and Collector (lb_api and lb_collector, respectively) then forwarded traffic from relevant front-end ports to these host-exposed backend ports. After host reboots, Docker would then expose different random ephemeral port numbers that now differed from the load balancer configurations, causing the Sysdig application to be unreachable post-reboot.
To address this issue, starting with version 890, the admin console now assigns static port numbers in lieu of the previous ephemeral numbers. This ensures the load balancer configurations are always in-sync with the exposed backend container port numbers. The default exposed ports are in an unassigned range and hence are unlikely to conflict with any other services you may be running on your hosts. However, if you do notice a conflict, immediately upon upgrade to version 890, expand the Advanced Settings in the Settings tab and locate the Exposed port options.
Replace the default values with your choice of unique, non-conflicting ports. When you click Save, you will be prompted to restart the Sysdig platform.