Add a lock_cluster command ?

[clintongormley]

Just an idea, how about being able to lock the cluster to prevent writes and or reads.

For instance, when the cluster is starting up and restoring from the gateway, it may take a few seconds to load all of the data. It'd be nice to say:

lock the cluster until cluster_health is green

Also, when changing a mapping, (thus forcing a reindex), we could do something like:

• generate new index
• lock cluster
• rename the indices
• flush index
• unlock cluster when cluster_health is green

ta

clint

[kimchy]

regarding the first point, the APIs will simply wait (until the timeout) if the cluster can't fullfil the requests yet.

An index level lock is interesting (mainly, I guess, to disallow dirty operations).

[clintongormley]

This seems to be handled nicely in recent releases