I have been trying to take a backup of a database and it failed with Internal failure ARC0001 :DBCHWAT and it said we had parcel but fetch code we got back on session -1 was 1 . On further investigation it was noted that it is due to untranslatable characters noted in the table.
This is a table with all VARCHAR columns with CHARACTER SET LATIN NOT CASESPECIFIC. We deleted the data with junk characters by identifying the rows that satsfied the below conditions:
DELETE FROM <database.tablename> WHERE TRANSLATE_CHK(<columnname> USING LATIN_TO_UNICODE) <> 0;
DELETE FROM <database.tablename> WHERE TRANSLATE_CHK(<columnname> USING LATIN_TO_KANJISJIS) <> 0;
Even after deleting the rows that had the junk characters we still noted database backup failing.
Please let me know if i am missing anything to identify untranslatable characters and remove those rows. Please help.
You verified also that you deleted successfully?
This link of translate_chk, you can see one by one:
Yes I did verify that all the rows with junk characters got deleted successfully. Is there anything else that i can perform to ensure that there are no junk characters left in the table.
Oops!Eusha, I should have suggested you to consult before deleting too. I have gone in your direction. The very basic idea of backup is that we can restore later. Data is very important, that is why we do backup. Please follow the recommendation of Tod.
We experienced this exact error message while taking full system dictionary backup.
System is running Teradata 14.10.02.08 and TTU version is 14.10.
I've opened a ticket with Teradata BAR team, will keep this thread updated with their response.
If DBCHWAT is preceded by a non-fatal error about STAT COLLECTION (which was in my case), here's a workaround for this error:
NTA 3561 : ARCHIVE may fail with 'ARC0729:SHOW Stat Collection query failed with error 6705 or 6706'
Fixed with DBS14.10.02.10 an higher
Use 'SKIP STAT COLLECTION' option in ARC (Available by JIRA ARC-8301 in ARC_22.214.171.124 and ARC_126.96.36.199)
Identify a CHAR column that has untranslatable character(s). If nothing can be done with those columns, then drop and collect new statistics on affected columns.