I’m running StarRocks 3.4.3 with a shared data backend.
I’ve run into a similar issue a couple of times while attempting a BROKER LOAD. Either immediately or after a few dozen seconds the job will fail with the error partition's index is not equal with schema.
The last time this happened, I was able to resubmit the job and after several retries it was successful in loading the data.
I’ve not been able to find any issues on GitHub or topics on this forum that address this, nor any mention of it in the documentation, leaving me not knowing where/how I’m supposed to troubleshoot or resolve this.
I can provide logs, etc., but am first interested in if this is an issue that has been encountered before and what the underlying cause is.
A little more detail is that this is a broker load job that is attempting to insert/update approximately 350,000 records into a primary key table; in this case it is almost all updates. The table is partitioned by an int field (org_id) and year. The source has data going back three years and the records in that entire timeframe were updated, so there are historical records to be updated, not just recent ones.
The full message from information_schema.loads is
type:LOAD_RUN_FAIL; msg:partition's index is not equal with schema's, part_index=15199, schema_index=24719: BE:10001. Retry again
Once, out of the many times the load job was attempted, I did also receive the error
type:LOAD_RUN_FAIL; msg:automatic create partition failed. error: txn 15373 is ABORTED: BE:10001