You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@accumulo.apache.org by Mike Miller <mm...@apache.org> on 2018/07/27 20:47:47 UTC

Bulk Import with bad Column Visibility

Found something interesting while helping a user find a file that was bulk
imported with a bad Column Visibility.  I was about to write it off as
"well if you ingest data with a bad CV then you won't be able to get it
back" but it looks like validation was added to prevent bulk import with
invalid CV in ACCUMULO-360[1].  Validation was added to
AccumuloFileOutputFormat but I don't see recent versions using
AccumuloFileOutputFormat
during bulk import.

I did some bulk imports with a CV of "A|B|" using Uno across different
versions:
1.6 - Rfile imported, throws server error on scan
1.7 - Rfile imported, scan only returns rows with valid visibility but does
not throw error
1.9 - Rfile imported, scan only returns rows with valid visibility but does
not throw error

I attached the stacktrace, which only shows up in 1.6.

Has anyone ran into this issue before?  Perhaps this validation was removed
for performance reasons?

[1] https://issues.apache.org/jira/browse/ACCUMULO-360