You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@drill.apache.org by Matt Keranen <hw...@gmail.com> on 2018/11/27 20:35:59 UTC

File "does not exist" error on non-distributed filesystem cluster

Have 4 nodes running drillbits version 1.14 for queries over JSON files in
the regular filesystem (not HDFS).

Each node has an identical directory structure, but not all file names
exist on all nodes, and any query in the form of "SELECT ... FROM
dfs.logs.`logs*.json.gz`" fails with:

    Error: DATA_READ ERROR: Failure reading JSON file - File
file:/localdata/logs/logs.xxx.json.gz does not exist

where the filename may change, but is always one that exists on some but
not all nodes.

Is there a configuration for Drill where drillbits querying non-distributed
filesystems don't expect all files to exist on all nodes?