This is an automated email from the ASF dual-hosted git repository.
cloud-fan pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/spark.git
The following commit(s) were added to refs/heads/master by this push:
new 433aee3f46b0 [SPARK-58094][CONNECT][FOLLOWUP] Fix flaky BindException
in SparkConnectServiceKeepAliveSuite
433aee3f46b0 is described below
commit 433aee3f46b0e598835e338477b820991981b9ad
Author: Vinod KC <[email protected]>
AuthorDate: Sat Jul 18 05:55:20 2026 +0800
[SPARK-58094][CONNECT][FOLLOWUP] Fix flaky BindException in
SparkConnectServiceKeepAliveSuite
### What changes were proposed in this pull request?
Three tests in `SparkConnectServiceKeepAliveSuite` each call
`SparkConnectService.stop()` immediately followed by
`SparkConnectService.start()` on the same port. The bare `stop()` uses
`server.shutdownNow()` which is non-blocking — the OS-level TCP socket may
still be bound when `start()` tries to bind, causing an intermittent
`BindException` with `maxRetries=0`.
The fix: replace the bare `stop()` calls in these tests with
`stop(Some(30), Some(TimeUnit.SECONDS))`. This takes the existing `if`-branch
in `stop()` that calls `server.shutdown()` followed by
`server.awaitTermination(30s)`, giving the server time to fully release the
port before returning. The production `stop()` code is unchanged.
### Why are the changes needed?
`SparkConnectServiceKeepAliveSuite` (introduced in #57202) was hitting
intermittent CI failures:
java.net.BindException: Failed to bind to address 0.0.0.0/0.0.0.0:15312:
Service 'org.apache.spark.sql.connect.service.SparkConnectService' failed
after 0 retries (starting from 15312)!
### Does this PR introduce _any_ user-facing change?
No. The change is test-only. Production code is unchanged.
### How was this patch tested?
The existing `SparkConnectServiceKeepAliveSuite` tests cover the stop/start
lifecycle. No new tests are needed — this fix eliminates the race those tests
were hitting. Scala and Python linters pass.
### Was this patch authored or co-authored using generative AI tooling?
Generated-by: Claude Sonnet 4.6
Closes #57342 from vinodkc/fix-spark58094-flaky-test.
Authored-by: Vinod KC <[email protected]>
Signed-off-by: Wenchen Fan <[email protected]>
---
.../connect/service/SparkConnectServiceKeepAliveSuite.scala | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git
a/sql/connect/server/src/test/scala/org/apache/spark/sql/connect/service/SparkConnectServiceKeepAliveSuite.scala
b/sql/connect/server/src/test/scala/org/apache/spark/sql/connect/service/SparkConnectServiceKeepAliveSuite.scala
index d1d273016e76..de8d6b84e30d 100644
---
a/sql/connect/server/src/test/scala/org/apache/spark/sql/connect/service/SparkConnectServiceKeepAliveSuite.scala
+++
b/sql/connect/server/src/test/scala/org/apache/spark/sql/connect/service/SparkConnectServiceKeepAliveSuite.scala
@@ -121,7 +121,7 @@ class SparkConnectServiceKeepAliveSuite extends
SparkConnectServerTest {
// behavior), this client's cadence would violate that coupled invariant;
with today's fixed,
// decoupled GRPC_KEEPALIVE_PERMIT_TIME_SECONDS floor (10s), it's
comfortably tolerated.
val clientKeepAliveMs =
(SparkConnectService.GRPC_KEEPALIVE_PERMIT_TIME_SECONDS + 1) * 1000
- SparkConnectService.stop()
+ SparkConnectService.stop(Some(30), Some(TimeUnit.SECONDS))
withSparkEnvConfs(
Connect.CONNECT_GRPC_BINDING_PORT.key -> serverPort.toString,
Connect.CONNECT_GRPC_KEEPALIVE_ENABLED.key -> "true",
@@ -156,7 +156,7 @@ class SparkConnectServiceKeepAliveSuite extends
SparkConnectServerTest {
client.shutdown()
}
} finally {
- SparkConnectService.stop()
+ SparkConnectService.stop(Some(30), Some(TimeUnit.SECONDS))
withSparkEnvConfs(
Connect.CONNECT_GRPC_BINDING_PORT.key -> serverPort.toString,
Connect.CONNECT_GRPC_KEEPALIVE_ENABLED.key -> "true",
@@ -190,7 +190,7 @@ class SparkConnectServiceKeepAliveSuite extends
SparkConnectServerTest {
// truly idle connection -- hence one call to establish the transport,
then real idle time
// (no further calls) spanning several ping intervals, then a
connection-count check.
val clientKeepAliveMs =
(SparkConnectService.GRPC_KEEPALIVE_PERMIT_TIME_SECONDS + 1) * 1000
- SparkConnectService.stop()
+ SparkConnectService.stop(Some(30), Some(TimeUnit.SECONDS))
withSparkEnvConfs(
Connect.CONNECT_GRPC_BINDING_PORT.key -> serverPort.toString,
Connect.CONNECT_GRPC_KEEPALIVE_ENABLED.key -> "false",
@@ -238,7 +238,7 @@ class SparkConnectServiceKeepAliveSuite extends
SparkConnectServerTest {
}
} finally {
relay.close()
- SparkConnectService.stop()
+ SparkConnectService.stop(Some(30), Some(TimeUnit.SECONDS))
withSparkEnvConfs(
Connect.CONNECT_GRPC_BINDING_PORT.key -> serverPort.toString,
Connect.CONNECT_GRPC_KEEPALIVE_ENABLED.key -> "true",
@@ -253,7 +253,7 @@ class SparkConnectServiceKeepAliveSuite extends
SparkConnectServerTest {
"SPARK-58094: disabling spark.connect.grpc.keepAlive.enabled reverts to
the pre-fix hang") {
// Restart the real service with keepalive fully disabled (not just given
short/aggressive
// timing) to prove the flag genuinely gates the fix rather than only
tuning its timing.
- SparkConnectService.stop()
+ SparkConnectService.stop(Some(30), Some(TimeUnit.SECONDS))
withSparkEnvConfs(
Connect.CONNECT_GRPC_BINDING_PORT.key -> serverPort.toString,
Connect.CONNECT_GRPC_KEEPALIVE_ENABLED.key -> "false",
@@ -321,7 +321,7 @@ class SparkConnectServiceKeepAliveSuite extends
SparkConnectServerTest {
}
} finally {
// Restore the enabled server for afterAll()/subsequent tests in this
suite.
- SparkConnectService.stop()
+ SparkConnectService.stop(Some(30), Some(TimeUnit.SECONDS))
withSparkEnvConfs(
Connect.CONNECT_GRPC_BINDING_PORT.key -> serverPort.toString,
Connect.CONNECT_GRPC_KEEPALIVE_ENABLED.key -> "true",
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
