Conversation
30703d2 to
eb1b4f7
Compare
plugin/trino-hive/src/test/resources/hive_s3_insert_overwrite/hive-core-site.xml
Outdated
Show resolved
Hide resolved
plugin/trino-iceberg/src/test/java/io/trino/plugin/iceberg/TestIcebergSplitSource.java
Outdated
Show resolved
Hide resolved
testing/trino-testing/src/main/java/io/trino/testing/AbstractTestDistributedQueries.java
Outdated
Show resolved
Hide resolved
testing/trino-testing/src/main/java/io/trino/testing/BaseConnectorTest.java
Outdated
Show resolved
Hide resolved
eb1b4f7 to
c2681d2
Compare
|
@findepi @findinpath I'm struggling to solve one failing test in newly added integration tests. From what I observed/debugged. Iceberg connector behave like this when performing CTAS operations for all use cases given by data provider.
|
|
I suspect we have an issue here The code above doesn't take into account timeout exceptions on the client side (configured to 10s). |
|
Thanks to @findinpath I got full understanding on cause. It's because HMS takes longer when acting with MinIO in test and we get timeout on Thrift |
5b81d7f to
c4ae0ec
Compare
7e8f15a to
70b05a8
Compare
testing/trino-testing/src/main/java/io/trino/testing/AbstractTestDistributedQueries.java
Outdated
Show resolved
Hide resolved
plugin/trino-iceberg/src/test/java/io/trino/plugin/iceberg/BaseIcebergMinioConnectorTest.java
Outdated
Show resolved
Hide resolved
plugin/trino-iceberg/src/test/java/io/trino/plugin/iceberg/BaseIcebergMinioConnectorTest.java
Outdated
Show resolved
Hide resolved
...rino-iceberg/src/test/java/io/trino/plugin/iceberg/TestIcebergMinioParquetConnectorTest.java
Outdated
Show resolved
Hide resolved
...rino-iceberg/src/test/java/io/trino/plugin/iceberg/TestIcebergMinioParquetConnectorTest.java
Outdated
Show resolved
Hide resolved
4312aec to
6244872
Compare
6244872 to
bf9a7ab
Compare
|
@findepi @findinpath In current state test are getting green. |
9dbd9ce to
efa6911
Compare
Yes looking at it. What i can see is that most consuming test's are:
That takes around 30m by them self Also Even if i would reduce now time it will hit again soon.
@findepi WDYT Top test run times: |
@losipiuk @arhimondr why are these so slow? @aczajkowski how much the new test classes, MinIO-related, add here? |
In general all MinIO related test takes But in there is Those are all MinIO test |
65b7114 to
f88d8ee
Compare
8334de9 to
f5575a3
Compare
|
Rebased on top of master to grab: #11142 |
12185e8 to
a8c2788
Compare
4a7deb0 to
a99c62f
Compare
|
@findepi I've verified that is still required. Tests become flaky without it. Once to MinIO test starts to run in parallel we get OOM. |
|
(i prefer #11219) |
|
when doing so, can you please change commit title to |
Yes working on that now. To sync base commits and resolve merge conflicts. |
|
Marked as draft so that i don't mistake this one and #11219 again. |
|
@findepi another conflicts. Rebased on master. |
|
Closing as #11219 got merged. |
Introduce MinIO based data lake smoke tests for iceberg connector.
Redundant with: #11219 (2 first commits are shared)
We will merge eider Integration or Smoke test TBD.