Retry DoXCache during redis cluster scaling

[wyxloading]

Recently we got failed to fetch the cache because EXEC was aborted by redis or connection closed by DoMultiCache during redis cluster scaling ( relate to #696 )

We believe that fail reason should be something relation to cluster scaling.
Check source below, seem like in TRANSACTION do commands MULTI, PTTL, GET/MGET, EXEC, then redis server return a error MOVED {slot} {ip:port} for PTTL command, and EXEC get a EXECABORT error. But driver only check with the EXEC command result.

https://github.com/redis/rueidis/blob/main/pipe.go#L1446-L1482

Seem like we can fix this easily, just handle previous error in TRANSACTION and let driver handle retry automaticlly.

We will test it ASAP and maybe file a PR later.

[rueian]

just handle previous error in TRANSACTION

Do you refer to the MOVED error you mentioned? I think it’s a good idea to replace the EXECABORT with the MOVED in this case.

[wyxloading]

Do you refer to the MOVED error you mentioned?

Yes, and ASK error too. Check PR #701