Proposal: could wait times not impact overall pipelinerun timeout? #8701
Description
Feature request
Recently, we've been building with Konflux and the server has been overloaded, so a 90 min build has been taking over 2h40. Sometimes due to the lack of pods, builds take over 3hrs, which is the current overall timeout.
Admittedly I could increase that to 4hrs to allow for high traffic days when everyone is stuck in a queue waiting for a pod to spin up, but then if a process is ACTUALLY stuck (eg., it OOMed but didn't crash the pod) then I'll be hogging resources from others for ANOTHER hour. So keeping the max build time LOW is better as it makes me a better citizen on the shared server farm.
Therefore I'm wondering if we could EXCLUDE the wait time from the overall timeout.
For example, if my build's pipelinerun ACTUALLY is only processing for 1 hr of total pod-running-time (combined across all the serial and parallel taskruns), but it takes 4hrs of real human time for that 1 hr of work to be accomplished (due to server load / resource limitations, then my build would still be able to complete without being cancelled at the 3hr mark.
Use case
See above. 1 hr build that takes 4 hrs to complete due to server load / pod startup wait times should not be cancelled after 3 hrs.