You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If there are n indices with m shards each, we first run n*m search calls to simply fetch the max sequence number
for each shard and then perform the searches which fetch docs from each shard based on last seen seq no. and max seq no.
Optimized behaviour proposed
Change first search request from each shard to fetch all docs in descending order of seq no with filter on `seqno >
That way we are fetching docs and also able to note the sequence number correctly within the same call and avoid the first batch of search calls which simply help determine the sequence number
The text was updated successfully, but these errors were encountered:
Current behaviour
n
indices withm
shards each, we first runn*m
search calls to simply fetch the max sequence numberfor each shard and then perform the searches which fetch docs from each shard based on last seen seq no. and max seq no.
Optimized behaviour proposed
Change first search request from each shard to fetch all docs in descending order of seq no with filter on `seqno >
That way we are fetching docs and also able to note the sequence number correctly within the same call and avoid the first batch of search calls which simply help determine the sequence number
The text was updated successfully, but these errors were encountered: