How cluster-node-timeout affects cluster rebalancing/resharding

[stav alfi]

When I have 5 master nodes without any replicas, and each of the masters has 5000+ keys, I get an error when im trying to send commands (multi/pipeline/lua) which are ONLY related to the same slot:

[WARNING] Node 127.0.0.1:7000 has slots in importing state (5468).
[WARNING] Node 127.0.0.1:7001 has slots in migrating state (5468).

I read that this should not happen.

After increasing  cluster-node-timeout from 5000 to 500000 (500s), I didn't see any error at all any more. 

`redis-cli --cluster reshard` and `redis-cli --cluster rebalance` works great while alot of requests are sent to the redis-cluster (same as above).

Question

1. Did I really solve the error? Or is it a coincidence? 

2. If I solved the problem, what is the downside of increasing the cluster-node-timeout?

3. If I solved the problem, can I assume uptime during adding nodes,removing nodes, resharing slots and rebalancing nodes on the cluster?