Open nchaulet opened 9 hours ago
It seems that upgrading a integration stream with a lot of backing indices could sometime fail because the response is too large.
We should avoid retrieving all backing indices (settings, and mappings) and only return the current one for the write index.
That PR does that.
I added a unit test to verify we only call GET /<indices> with the current write index and not the datastream name
GET /<indices>
Pinging @elastic/fleet (Team:Fleet)
✅ unchanged
cc @nchaulet
Summary
It seems that upgrading a integration stream with a lot of backing indices could sometime fail because the response is too large.
We should avoid retrieving all backing indices (settings, and mappings) and only return the current one for the write index.
That PR does that.
Test
I added a unit test to verify we only call
GET /<indices>
with the current write index and not the datastream name