This feature is used to fail over between multiple redundant endpoints that can each serve your workload - designed for an
Active-Active (geo-distributed, bidirectionally-replicated) deployment, but usable over any set of endpoints, including ones
with no replication between them (see the full docs for what that means for your data). It is functionally similar to the
“multi-DB client” concept in other Redis clients (Jedis’ MultiDbClient, redis-py’s MultiDBClient); please see
full docs.
This is a shipped, supported feature, intended for production use. It is not a preview, it is not gated on an unreleased server version, and it does not depend on any server API that is subject to change. Bugs in it are treated as bugs, and can be reported and addressed in the usual way. No API changes are currently planned.
The [Experimental] marker is nonetheless a real one: while it is present, we reserve the right to adjust this API surface
without the usual backwards-compatibility guarantees. That is the whole reason it is here. Client-side geographic failover
is a large, brand-new public API arriving in a single release, and once a surface that size is locked down it is locked down
for good - so we would rather keep room to correct a design wrinkle found in real-world use than commit to every detail
on day one.
The marker is expected to be short-lived, and to be removed in a subsequent 3.1.x release once the design has had some exposure. At that point the diagnostic simply stops being emitted, with no code change needed on your side.
Suppressing the diagnostic is the normal way to use the feature; doing so acknowledges the above.
To suppress this message, add the following to your csproj file:
<NoWarn>$(NoWarn);SER007</NoWarn>
or more granularly / locally in C#:
#pragma warning disable SER007