[hotfix] Try to stabilize the flaky integration tests

[RocMarshal]

No description provided.

[yuanoOo]

@RocMarshal Hi there,

To fix the GitHub CI failure caused by the OceanBase container startup issue, I suggest the following:

1. Adjust the container startup parameters:

   .withMode(OceanBaseCEContainer.Mode.MINI)
   .withEnv("OB_DATAFILE_SIZE", "2G")
   .withEnv("OB_LOG_DISK_SIZE", "4G")
   .withStartupTimeout(Duration.ofMinutes(4));

   See the complete example here: https://github.com/oceanbase/flink-connector-oceanbase/blob/4b4f33376a6f60e3f5f3586a6d9904c1c356b8a5/flink-connector-oceanbase-base/src/test/java/com/oceanbase/connector/flink/OceanBaseMySQLTestBase.java#L62-L72

2. Upgrade the image version to 4.3.5-lts

[RocMarshal]

Hi, Thank you @yuanoOo @whhe for your help .

https://github.com/apache/flink-connector-jdbc/actions/runs/20158946377/job/57867060428?pr=184

It seems this issue has still occurred during this test.

Since I'm not very familiar with this database now, would you please provide some other suggestions? I'd be very grateful.

[yuanoOo]

Hi, Thank you @yuanoOo @whhe for your help .

https://github.com/apache/flink-connector-jdbc/actions/runs/20158946377/job/57867060428?pr=184

It seems this issue has still occurred during this test.

Since I'm not very familiar with this database now, would you please provide some other suggestions? I'd be very grateful.

Hi, try this version: oceanbase/oceanbase-ce:4.3.3.1-101000012024102216 . I didn't encounter this issue in the PR I just ran: oceanbase/spark-connector-oceanbase#86

[whhe]

A possible relevant information is that OceanBase startup time will be longer when available machine resources (such as memory and disk) are limited. Therefore, if the above suggestions still don't solve the problem, perhaps we can split the tests to different GHA runners, or using a lighter-weight image like seekdb.

[RocMarshal]

Thanks @yuanoOo @whhe for your help.
And I'll to push forward the next work based on your suggestions!