quic: increase timeout and keep alive

[0x0ece]

Problem

Quic timeout and keep alive are very low.
When a validator is not leader, it still has thousands of connections open that are not expected to send any data.
Because keep alive is 1s, each client sends a quic ping every second, which is a lot of unnecessary traffic.

Summary of Changes

Increased timeout to 60s and keep alive to 45s.

[lijunwangs]

Maybe we should different timeout values: stream timeout and idle connection timeout (controlled by the ping)

[lijunwangs]

We need to investigate why the CI is consistently failing for this PR. I am updating the branch again.

[behzadnouri]

@0x0ece Looks like this change has made test_snapshots_restart_validity very difficult to pass.
See "6/10 local-cluster" which takes 9 retries to pass: https://buildkite.com/anza/agave/builds/17947
(on another PR I didn't have any luck with many retries but the CI passed once I reverted this change).

The CI for this change is green but I am guessing the combination of this change with another PR (maybe #4586 ) is causing the test to fail. I have confirmed though reverting this change resolves the issue: https://buildkite.com/anza/agave/builds/17958

This is currently blocking work because the CI is not usable. Can we please revert this and test the change again?

[lijunwangs]

@0x0ece Looks like this change has made test_snapshots_restart_validity very difficult to pass. See "6/10 local-cluster" which takes 9 retries to pass: https://buildkite.com/anza/agave/builds/17947 (on another PR I didn't have any luck with many retries but the CI passed once I reverted this change).

The CI for this change is green but I am guessing the combination of this change with another PR (maybe #4586 ) is causing the test to fail. I have confirmed though reverting this change resolves the issue: https://buildkite.com/anza/agave/builds/17958

This is currently blocking work because the CI is not usable. Can we please revert this and test the change again?

okay. We can revert it. I will revert immediately @behzadnouri

[0x0ece]

Ok, definitely let’s investigate.
FYI, after this I collected more data. In a TPU session we’re getting something like 1m packets with txs and 2.3m pings from peers that are just idle, not sending txs.

[lijunwangs]

Ok, definitely let’s investigate. FYI, after this I collected more data. In a TPU session we’re getting something like 1m packets with txs and 2.3m pings from peers that are just idle, not sending txs.

PR for revert: #4637