It seems the monitor view page in Synthetics calls the API GET /internal/synthetics/pings to get the "recent" pings and it looks at moment().subtract(30, 'days') by default (here)
If old data lands on slow data tiers (e.g. frozen), it can take several seconds to run.
From the inspect, the query seems to run 2 times even if the body request is unchanged (few milliseconds of difference in the time range).
Describe a specific use case for the enhancement:
Whenever possible, attempt to avoid querying long time in the past and to it only once.
Users might use aggressive ILM policies which make data move to the frozen tier and queries might be slower than expected.
Maybe the graphs might offer a weekly window and offer to expand it further to monthly.
Describe the enhancement:
It seems the monitor view page in Synthetics calls the API
GET /internal/synthetics/pings
to get the "recent" pings and it looks atmoment().subtract(30, 'days')
by default (here)If old data lands on slow data tiers (e.g. frozen), it can take several seconds to run. From the inspect, the query seems to run 2 times even if the body request is unchanged (few milliseconds of difference in the time range).
Describe a specific use case for the enhancement:
Whenever possible, attempt to avoid querying long time in the past and to it only once.
Users might use aggressive ILM policies which make data move to the frozen tier and queries might be slower than expected.
Maybe the graphs might offer a weekly window and offer to expand it further to monthly.
Side note: https://github.com/elastic/kibana/issues/190559 should help avoiding querying the frozen tier globally