Open
Description
What happened?
Running sys-export on a larger cluster seems to stall and make the node co-ordinating the queries un-responsive. Looking at the created report files it is not clear where it exactly got stuck. CTRL+C did not terminate the command, had to fire a kill in the shell to stop the python process.
To discuss/Implement
- THROTTELING: Think about a possible throttling, especially for clusters with a large amount of jobs, shards, ...
- TIMEOUT: set an explicit statement timeout to avoid long running, blocking queries (althought not 100% whether that also works for queries on sys tables
- SHARDS: for troubleshooting purpose it might suffice to report the not started shards, same for the sys.allocations query. In case we fancy to have to complete list of shards, this could be moved to a cli arg, where the default is to report only not started shards.