Marco Pracucci
Marco Pracucci
I'm also wondering if we could improve the remote write protocol to pass some context information, like the min/max scrape interval.
> by first marking itself as not ready This should be already covered by `-shutdown-delay`: > How long to wait between SIGTERM and shutdown. After receiving SIGTERM, Mimir will report...
> flushing in-flight queries, what else @pracucci? Just to be on the same page, we shouldn't flush in-flight queries but wait until they're done. This in addition to `-shutdown-delay` should...
> in-flight queries but wait until they're done I checked the code and, _if I'm not missing anything_, I agree with you that we currently don't wait until in-flight queries...
> Though, I have the question of whether the query-frontend should wait even longer than -shutdown-delay, i.e. until all in-flight queries have finished? I think so. We don't want a...
Please also remember to sign the CLA. Thanks!
@kovrus Any update on this PR?
> I'm wondering whether it makes sense to modify github.com/thanos-io/objstore/providers/filesystem.Bucket to respect context cancellations, which would remove the need for cancellationRespectingBucket here and also bring the implementation in line with...
## How the store-gateway send `Series()` response to the client (querier) The store-gateway `Series()` gRPC endpoint is based on a stream. Each message sent on the gRPC stream contains a...
## Theory: HTTP2 flow control ### What is flow control gRPC is based on HTTP2, and HTTP2 has a flow control mechanism. The flow control is based on a window...