MOVED error when trying to read record that was created about the same time

[snbuch]

In my application that uses latest redis (v9) cluster, the active writes the record while the standby reads the same record around the same time. This works most of the time but run into MOVED error sometimes. This does not resolve in the 3 MaxRedirects but does work if retried after a short delay.

I am curious why redis uses LazyReload() rather than Reload() on this line:

https://github.com/redis/go-redis/blame/21bd40a47e56e61c0598ea1bdf8e02e67d1aa651/osscluster.go#L964

Can anyone clarify why LazyReload() is being used here?

Any ideas why the retries in the for loop don't resolve the MOVED error?

Has anyone else run into this issue?
Edit: It seems the MOVED error typically happens right after the initial deployment but very rare subsequently.