Closed dominiqueclarke closed 1 year ago
Pinging @elastic/uptime (Team:uptime)
We have made various other improvements where we can, but will keep this open to track the platform change to improve streaming responses (moving to the 8.6 board as we’ll be closing the 8.5 board down soon).
This is not valid anymore, so closing. Since we are moving away from streaming in new api structure
Related to https://github.com/elastic/cloud/issues/106308
Kibana version: 8.4.0
Original install method (e.g. download page, yum, from source, etc.): Cloud console
Description of the problem including expected versus actual behavior:
The project monitors api allows our users to store monitor configuration in source control and push them directly to Kibana, via the synthetics-agent, to generate monitors that can run either through our globally managed testing infrastructure, or via
elastic-agent
and private locations.When configuration monitors to run via
elastic-agent
via private locations in particular, the operations are slow.This causes timeouts when pushing even a small number of monitors, say 4 or 5, as the synthetics-agent has a 30-second timeout.
To resolve this, we use streaming to provide updates of each monitor operation as they happen. This works fine locally, but in Cloud the streaming response headers are not being honored, causing the entire response to be buffered and then sent back to the synthetics-agent. This buffering often takes longer than 30 seconds in even the smallest cases, causing timeouts.