Error on parallel load: SCIDB_LE_CHUNK_ALREADY_EXISTS


#1

Hi

I attempted to do a parallel load and I got this error message:

bash $ date; time iquery -naq "load(raw, '/local/scidb/test.scidb', -1)"; date       
Tue Dec  4 15:36:33 EST 2012
UserException in file: src/array/Array.cpp function: merge line: 314
Error id: scidb::SCIDB_SE_MERGE::SCIDB_LE_CHUNK_ALREADY_EXISTS
Error description: Merge error. An existing chunk cannot be updated.
Failed query id: 1100885561203

real    0m1.007s
user    0m0.009s
sys     0m0.005s
Tue Dec  4 15:36:34 EST 2012

I initially tried it on a system running SciDB 12.3, and then repeated the test on SciDB 12.10. The later system was Ubuntu 12.04.

I prepped my file using csv2scidb, specifying the chunk size as 100000 since I have 11 attributes (on average half the attributes are null). Matching chunk size in the raw array I am loading to.

Any thoughts? Thanks,
Dave