search

nck-2 / test-rep

0 stars 0 forks source link

OPTIMIZE for for RT in cluster #1292

Closed githubmanticore closed 1 year ago

githubmanticore commented 1 year ago

OPTIMIZE statement should be refactored for RT index in cluster as it works at each node and produce different disk chunks at each node.

Then different node will be chose as donor and that node SST index files and as disk chunks are different between various donor nodes and joiner node that cause that joiner always copy index files over network.

Maybe it is worth to send cluster wide event between merge of each disk chunks that each node will wait other nodes in cluster and merges same disk chunks in same sequence.

githubmanticore commented 1 year ago

➤ Stan commented:

still actual