Conversation
|
✅ No conflicts with other open PRs targeting |
|
| return block, err | ||
| }, time.Minute*2, time.Second*2).Should(gomega.BeNumerically(">", preRestartBlock), | ||
| "Node %d should be processing new blocks after restart", i) | ||
| l.Info().Msgf("Node %d is ready after upgrade", i) |
There was a problem hiding this comment.
but this chain client has nothing to do with the node... it's a standalone Ethereum client that connects to the RPC
There was a problem hiding this comment.
how else would you remove the sleep?
There was a problem hiding this comment.
just delete it or check something meaningful, i.e. node's health endpoints and not whether the blockchain has advanced beyond some block old block, because that has nothing to do with the node
Tofel
left a comment
Tofel
left a comment
There was a problem hiding this comment.
a.ChainClient has nothing to do with the node... it's a standalone Ethereum client that connects to the RPC. It is literally a chain client, and chain is not a shortcut for chainlink.
4 participants
After restarting nodes during a rolling upgrade, the test used a fixed
10-second sleep before checking upkeep counters. This is fragile: nodes
may take longer to become ready, wasting CI time, or may be ready sooner,
making the test unnecessarily slow.
Replace the sleep with an
Eventuallyblock that polls the chain clientto confirm the node has processed new blocks, indicating readiness. The
polling uses a 2-minute timeout with 2-second intervals, which is
consistent with the existing
UpkeepExecutionTimeoutpattern in the samefile.
cre-4227