Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
3.0.0-alpha1
-
None
-
None
Description
While investigating issues in HADOOP-15059 I ran the dtutil append command like this:
$ hadoop dtutil append -format protobuf foo foo.pb
expecting the append command to translate the existing tokens in file foo into the currently non-existent file foo.pb. Instead the command executed without error and overwrote foo instead of creating foo.pb as I expected. I now understand how append works, but it was very surprising to have dtutil silently ignore filenames requested on the command-line. At best it is a bit surprising to the user. At worst it clobbers data the user did not expect to be overwritten.