This should make it so the session client is a bit more robust and long running jobs are less likely to fail in the middle of the night with no one to restart them. It may not always work, but it should help
Questions
if a request to one host fails, should all other requests to the host pause or should this be configurable on a request basis? if so How?
Should it be opt in on a request basis? (similar to the cache stuff)
This should make it so the session client is a bit more robust and long running jobs are less likely to fail in the middle of the night with no one to restart them. It may not always work, but it should help
Questions