Open tarzanek opened 1 year ago
it would make usability much better if Scylla Manager can somehow properly watch errors from L&S
L&S is called via REST API. The only error we can observe depends on what Scylla may return for the call. Manager is handling these errors (logging them).
and of course have support for some timeout so this won't be "stuck" forever
This is risky. We need to define "forever".
What we can do is to add restore config flag defining the l&s timeout, but what to do if we hit the timeout ? We cannot red-o the operation. Should we fail the restore ?
hi guys it would make usability much better if Scylla Manager can somehow properly watch errors from L&S and of course have support for some timeout so this won't be "stuck" forever
Force Scylla Core to get manager for L&S ala https://github.com/scylladb/scylladb/pull/12661 ?
We want to guard against https://github.com/scylladb/scylla-enterprise/issues/2849 and similar ( https://github.com/scylladb/scylla-enterprise/issues/2724 )
@karol-kokoszka fyi