Crate write behaviour -
i have question on crate write behavior.
from crate documentation
one reason replicas written synchronously. making write task wait replicas written somewhere in data center hundreds of miles away can lead noticeable latency , cause cluster slow down.
if have cluster of 3 nodes , if 1 node down, behavior when insert data? fail or crate sync data node down when up?
regards albin
as written here: https://crate.io/docs/en/latest/storage_consistency.html?highlight=quorum
write operations handled differently reads. such operations synchronous on active replicas following flow:
the primary shard , active replicas looked in cluster state given operation. primary shard , quorum of configured replicas need available step succeed.
...
the default quorum is:
int( (primary + number_of_replicas) / 2 ) + 1
so not fail long quorum achieved. after failed node operate normal again, crate take care create replicas again on node if replica not created on running node.
Comments
Post a Comment