Guillaume Nodet created CAMEL-24018:
---------------------------------------
Summary: Fix flaky OpensearchIndexIT - disk usage exceeded
flood-stage watermark
Key: CAMEL-24018
URL: https://issues.apache.org/jira/browse/CAMEL-24018
Project: Camel
Issue Type: Bug
Components: camel-opensearch
Reporter: Guillaume Nodet
The {{OpensearchIndexIT}} integration test is flaky on CI. Over the last 12
days, it shows 7 failed runs and 7 flaky runs out of 215 total runs.
All 5 test methods in the class fail with the same root cause:
{code}
cluster_block_exception: index [twitter] blocked by: [TOO_MANY_REQUESTS/12/disk
usage exceeded flood-stage watermark, index has read-only-allow-delete block]
{code}
This happens because the OpenSearch test container disk fills up on CI machines
with limited disk space. When disk usage exceeds the flood-stage watermark,
OpenSearch puts all indices into read-only-allow-delete mode, causing all write
operations to fail with 429 Too Many Requests.
The fix has two parts:
# Disable disk allocation watermarks in the OpenSearch test container
configuration so the container never blocks writes due to disk pressure.
# Add {{@AfterEach}} cleanup in {{OpensearchIndexIT}} to delete the "twitter"
index after each test, preventing data accumulation across tests.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
