Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP

Loading…

Peer Recovery: Batch translog based operation recovery based on size and not just number of operations #729

Closed
kimchy opened this Issue · 1 comment

1 participant

@kimchy
Owner

Currently, the peer recovery of translog does batching based on operation (defaults to 100). Batching based on size as well is important since there might be big docs that will prove problematic to recover.

The following parameters allow to control peer recovery:

  • index.shard.recovery.file_chunk_size: The chunks index files will be broken down to when recovered. Defaults to 100kb.
  • index.shard.recovery.translog_size: The batch size of translog operations. Defaults to 100kb.
  • index.shard.recovery.translog_ops: The number of operations per batch. Defaults to 1000.
@kimchy
Owner

eer Recovery: Batch translog based operation recovery based on size and not just number of operations, closed by 6097365.

@deinspanjer deinspanjer referenced this issue from a commit
@kimchy kimchy eer Recovery: Batch translog based operation recovery based on size a…
…nd not just number of operations, closes #729.
6097365
This issue was closed.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.