Repository navigation
feat(inkless:storage): default S3 API call timeouts to 2s and 1s - #791
Merged
Merged
Conversation
Fail hung S3 calls sooner instead of relying on the AWS SDK's unbounded defaults, matching the GCS connect and read timeout defaults. Co-authored-by: Cursor <cursoragent@cursor.com>
Give the server-error metrics integration test enough time to exhaust adaptive retries. This keeps the test focused on S3 response classification while production keeps the new fail-fast defaults. Co-authored-by: Cursor <cursoragent@cursor.com>
jeqo
approved these changes
Sep 9, 2026
jeqo
pushed a commit
that referenced
this pull request
Sep 9, 2026
* feat(inkless:storage): default S3 API call timeouts to 2s and 1s Fail hung S3 calls sooner instead of relying on the AWS SDK's unbounded defaults, matching the GCS connect and read timeout defaults. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(inkless:storage): isolate S3 error metrics timeout Give the server-error metrics integration test enough time to exhaust adaptive retries. This keeps the test focused on S3 response classification while production keeps the new fail-fast defaults. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com>
jeqo
pushed a commit
that referenced
this pull request
Sep 9, 2026
* feat(inkless:storage): default S3 API call timeouts to 2s and 1s Fail hung S3 calls sooner instead of relying on the AWS SDK's unbounded defaults, matching the GCS connect and read timeout defaults. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(inkless:storage): isolate S3 error metrics timeout Give the server-error metrics integration test enough time to exhaust adaptive retries. This keeps the test focused on S3 response classification while production keeps the new fail-fast defaults. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com>
jeqo
pushed a commit
that referenced
this pull request
Sep 9, 2026
* feat(inkless:storage): default S3 API call timeouts to 2s and 1s Fail hung S3 calls sooner instead of relying on the AWS SDK's unbounded defaults, matching the GCS connect and read timeout defaults. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(inkless:storage): isolate S3 error metrics timeout Give the server-error metrics integration test enough time to exhaust adaptive retries. This keeps the test focused on S3 response classification while production keeps the new fail-fast defaults. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com>
viktorsomogyi
pushed a commit
that referenced
this pull request
Sep 30, 2026
… switch systests (#813) * fix(inkless:test): raise S3 timeouts and avoid the isolated broker in switch systests Six `InklessClassicToDisklessSwitchTest` cases fail on the nightly run for two reasons unrelated to the switch itself. #791 set the S3 client defaults to a 2 s call timeout and a 1 s attempt timeout. The degraded-network tests throttle every broker NIC, including the path to MinIO, to 2 Mbit/s with 200 ms latency, so WAL uploads outlast 2 s and fail with `ApiCallTimeoutException`. The producer retries each failed produce, so it either never recovers or the consumer reads more records than were acked. `_create_kafka` now sets the call timeout to 60 s and the attempt timeout to 30 s. That fixes `test_switch_follower_crash_with_lagged_hw`, `test_switch_leader_crash` (both variants), `test_switch_leader_restart`, and `test_switch_rolling_restart`, which have failed since #791 merged. `test_switch_follower_network_partition` isolates a follower and then runs `kafka-configs.sh` from `self.nodes[0]`. When that node is the isolated follower, the admin client can't reach its peers, and the describe that precedes the alter hits its 30 s limit. `_switch_topic_to_diskless` now takes a `node`, and both network-partition tests pass one from the reachable side. `test_switch_leader_network_partition` has the same latent bug, since the leader can also be `self.nodes[0]`. ## Test plan Run the Inkless System Tests workflow on this branch and check that the six cases pass. `test_switched_topic_urp_clears_after_replica_recovery` still fails: that failure is the stranded-follower bug fixed by #760. Co-authored-by: Cursor <cursoragent@cursor.com> * Fix comment formatting in inkless_topic_switch_test.py Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
jeqo
pushed a commit
that referenced
this pull request
Oct 19, 2026
… switch systests (#813) * fix(inkless:test): raise S3 timeouts and avoid the isolated broker in switch systests Six `InklessClassicToDisklessSwitchTest` cases fail on the nightly run for two reasons unrelated to the switch itself. #791 set the S3 client defaults to a 2 s call timeout and a 1 s attempt timeout. The degraded-network tests throttle every broker NIC, including the path to MinIO, to 2 Mbit/s with 200 ms latency, so WAL upload 8000 s outlast 2 s and fail with `ApiCallTimeoutException`. The producer retries each failed produce, so it either never recovers or the consumer reads more records than were acked. `_create_kafka` now sets the call timeout to 60 s and the attempt timeout to 30 s. That fixes `test_switch_follower_crash_with_lagged_hw`, `test_switch_leader_crash` (both variants), `test_switch_leader_restart`, and `test_switch_rolling_restart`, which have failed since #791 merged. `test_switch_follower_network_partition` isolates a follower and then runs `kafka-configs.sh` from `self.nodes[0]`. When that node is the isolated follower, the admin client can't reach its peers, and the describe that precedes the alter hits its 30 s limit. `_switch_topic_to_diskless` now takes a `node`, and both network-partition tests pass one from the reachable side. `test_switch_leader_network_partition` has the same latent bug, since the leader can also be `self.nodes[0]`. ## Test plan Run the Inkless System Tests workflow on this branch and check that the six cases pass. `test_switched_topic_urp_clears_after_replica_recovery` still fails: that failure is the stranded-follower bug fixed by #760. Co-authored-by: Cursor <cursoragent@cursor.com> * Fix comment formatting in inkless_topic_switch_test.py Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
jeqo
pushed a commit
that referenced
this pull request
Oct 19, 2026
… switch systests (#813) * fix(inkless:test): raise S3 timeouts and avoid the isolated broker in switch systests Six `InklessClassicToDisklessSwitchTest` cases fail on the nightly run for two reasons unrelated to the switch itself. #791 set the S3 client defaults to a 2 s call timeout and a 1 s attempt timeout. The degraded-network tests throttle every broker NIC, including the path to MinIO, to 2 Mbit/s with 200 ms latency, so WAL uploads outlast 2 s and fail with `ApiCallTimeoutException`. The producer retries each failed produce, so it either never recovers or the consumer reads more records than were acked. `_create_kafka` now sets the call timeout to 60 s and the attempt timeout to 30 s. That fixes `test_switch_follower_crash_with_lagged_hw`, `test_switch_leader_crash` (both variants), `test_switch_leader_restart`, and `test_switch_rolling_restart`, which have failed since #791 merged. `test_switch_follower_network_partition` isolates a follower and then runs `kafka-configs.sh` from `self.nodes[0]`. When that node is the isolated follower, the admin client can't reach its peers, and the describe that precedes the alter hits its 30 s limit. `_switch_topic_to_diskless` now takes a `node`, and both network-partition tests pass one from the reachable side. `test_switch_leader_network_partition` has the same latent bug, since the leader can also be `self.nodes[0]`. ## Test plan Run the Inkless System Tests workflow on this branch and check that the six cases pass. `test_switched_topic_urp_clears_after_replica_recovery` still fails: that failure is the stranded-follower bug fixed by #760. Co-authored-by: Cursor <cursoragent@cursor.com> * Fix comment formatting in inkless_topic_switch_test.py Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
jeqo
pushed a commit
that referenced
this pull request
Oct 19, 2026
… switch systests (#813) * fix(inkless:test): raise S3 timeouts and avoid the isolated broker in switch systests Six `InklessClassicToDisklessSwitchTest` cases fail on the nightly run for two reasons unrelated to the switch itself. #791 set the S3 client defaults to a 2 s call timeout and a 1 s attempt timeout. The degraded-network tests throttle every broker NIC, including the path to MinIO, to 2 Mbit/s with 200 ms latency, so WAL uploads outlast 2 s and fail with `ApiCallTimeoutException`. The producer retries each failed produce, so it either never recovers or the consumer reads more records than were acked. `_create_kafka` now sets the call timeout to 60 s and the attempt timeout to 30 s. That fixes `test_switch_follower_crash_with_lagged_hw`, `test_switch_leader_crash` (both variants), `test_switch_leader_restart`, and `test_switch_rolling_restart`, which have failed since #791 merged. `test_switch_follower_network_partition` isolates a follower and then runs `kafka-configs.sh` from `self.nodes[0]`. When that node is the isolated follower, the admin client can't reach its peers, and the describe that precedes the alter hits its 30 s limit. `_switch_topic_to_diskless` now takes a `node`, and both network-partition tests pass one from the reachable side. `test_switch_leader_network_partition` has the same latent bug, since the leader can also be `self.nodes[0]`. ## Test plan Run the Inkless System Tests workflow on this branch and check that the six cases pass. `test_switched_topic_urp_clears_after_replica_recovery` still fails: that failure is the stranded-follower bug fixed by #760. Co-authored-by: Cursor <cursoragent@cursor.com> * Fix comment formatting in inkless_topic_switch_test.py Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Summary
inkless.storage.s3.api.call.timeoutto2000ms (full call, including retries) andinkless.storage.s3.api.call.attempt.timeoutto1000ms (single attempt).null, so the AWS SDK applied its own (much longer) timeouts. The new defaults fail hung S3 calls sooner and align with the GCS connect and read timeout defaults.Test plan
./gradlew :storage:inkless:test --tests io.aiven.inkless.storage_backend.s3.S3StorageConfigTest --tests io.aiven.inkless.storage_backend.s3.S3ClientBuilderTestMade with Cursor