Repository navigation
Object name contains unsupported characters #7678
Description
Apache Iceberg version
1.2.1 (latest release)
Query engine
Spark
Please describe the bug 🐞
Some weird stuff, I tried to run:
DELETE FROM default.test_positional_mor_deletes WHERE number IN (1, 3, 5, 7, 9, 11)And it failed:
23/05/22 12:15:17 WARN TaskSetManager: Lost task 0.0 in stage 25.0 (TID 32) (0436561124c6 executor driver): software.amazon.awssdk.services.s3.model.S3Exception: Object name contains unsupported characters. (Service: S3, Status Code: 400, Request ID: 17617637CB486230, Extended Request ID: e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855)
Turns out the path has a double slash:
s3://warehouse/default/test_positional_mor_deletes/data//00000-32-70be11f7-3c4b-40e0-b35a-334e97ef6554-00001-deletes.parquet
When I run:
DELETE FROM default.test_positional_mor_deletes WHERE number = 1 Everything is fine 😨 I see the error as well when using number % 2 = 0 (what I tried initially).
Activity
Fokko commentedon May 22, 2023
The table location looks good.
I think it has to do with the minio configuraiton
Spark: Positional deletes creates unpartitioned path
edgarRd commentedon May 24, 2023
I wonder if this is similar to what I saw in #7635 step 4.
Fokko commentedon May 24, 2023
Hey @edgarRd that is correct. I ran your steps in the issue, and was able to replicate it. This will fix step 4.
Spark: Positional deletes creates unpartitioned path