fix: start new connections independently of connection failures #5927
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
With current implementation
every time connection fails
we take the next delay from
delays
iterator.In the worst case first 4 DNS results
immediately refuse connection
and we start fifth connection attempt
with 1 year timeout,
effectively continuing all remaining
connection attempts without concurrency.
With new implementation
new connection attempts are
added to
connection_attempt_set
independently of connection failures
and after 10 seconds
we always end up with five
parallel connection attempts
as long as there are enough IP addresses.