There are 2 ways to handle HA in redis - sentinel and redis cluster.
Sentinel
If a master is not working as expected, Sentinel can start a failover
process where a slave is promoted to master, the other additional
slaves are reconfigured to use the new master, and the applications
using the Redis server informed about the new address to use when
connecting
Q: Since there might be connectivity issue I would like to configure sentinels in order to allow brain split for high availability
This is an anti-pattern of using sentinel. Here’s a similar example with even number of nodes explained in the docs
Example 1: just two Sentinels, DON'T DO THIS
In the above configuration we created two masters (assuming S2 could
failover without authorization) in a perfectly symmetrical way.
Clients may write indefinitely to both sides, and there is no way to
understand when the partition heals what configuration is the right
one, in order to prevent a permanent split brain condition. So please
deploy at least three Sentinels in three different boxes always.
Q: Now the question what will happen after connectivity of subnets is restored. Sentinels would detect brain split and two masters?
This data will be lost forever since when the partition will heal, the
master will be reconfigured as a slave of the new master, discarding
its data set.
Q: Data from survived master would be pushed to a downgraded master and he would need to drop all data diff which was gained during connectivity issue?
Yes
Q: Can I configure something in order for data to be merged?
You can't, redis would never merge anything
Redis cluster
What is this beast for?
The ability to automatically split your dataset among multiple nodes.
The ability to continue operations when a subset of the nodes are
experiencing failures or are unable to communicate with the rest of
the cluster.
So it’s basically a multiple writer solution. But it doesn’t support merge operations either
Redis Cluster design avoids conflicting versions of the same key-value
pair in multiple nodes as in the case of the Redis data model this is
not always desirable. Values in Redis are often very large; it is
common to see lists or sorted sets with millions of elements. Also
data types are semantically complex. Transferring and merging these
kind of values can be a major bottleneck and/or may require the
non-trivial involvement of application-side logic, additional memory
to store meta-data, and so forth.
Back to your scenario
Quoting from here
Fundamental things to know about Sentinel before deploying
You need at least three Sentinel instances for a robust deployment.
The three Sentinel instances should be placed into computers or
virtual machines that are believed to fail in an independent way. So
for example different physical servers or Virtual Machines executed on
different availability zones.
Note you can place sentinels on client machines too - this approach is heavily used in redis demos https://redis.io/topics/sentinel
You can also go with the cluster solution but it's harder to configure it + it has some limitations on multi-key operations and you'll still need to provide majority of nodes in case one of the subnets goes down to have some sort of HA
