Alternative implementation of abnormal machine stop and container events propagation #7653
Labels
kind/task
Internal things, technical debt, and to-do tasks to be performed.
lifecycle/stale
Denotes an issue or PR has remained open with no activity and has become stale.
severity/P2
Has a minor but important impact to the usage or development of the system.
Milestone
Description
Since we have changed the usage of
OpenShiftClient
from one instance per action to one instance per Che server in the scope of PR, we have one connection pool which has limits for a number of idled connections (it equals to 5 connections by default but it can be overridden). Based on this we have no way to watch pod actions and container events for each workspace with using 2 separate WebSocket connections because at some point we overload the connection pool.It is possible to use one websocket connection and listen to events of all workspaces if the Che server is configured in the following ways:
But there is no clear way how to implement this when Che server uses users' token (like OSIO scenario) or credentials of a user which doesn't have system admin rights.
So it is needed to investigate possible ways how to listen to OpenShift events which will work for all configuration combinations.
The text was updated successfully, but these errors were encountered: