I decided to embark on a short journey into querying drill profiles to get interesting statistics from a shared cluster (where I have root priv's)
What I noticed was that if I attempt to query on the 'user' field within the JSON file (which appears as the element), that drill returns to me the user who I am currently impersonating...not the user who originally ran the query: (drill 1.0) /usr/bin/sqlline -u jdbc:drill: -n root then: select user from `profiles/2aa32e9e-bdae-8949-8461-c14dafe63ee0.sys.drill` ; +-------+ | user | +-------+ | *root* | +-------+ cat 2aa32e9e-bdae-8949-8461-c14dafe63ee0.sys.drill|egrep -o 'user\":\"[a-z]+\"' user":"apernsteiner" Now, I stumbled across a workaround, which is to alias the table (t) and prefix the 'user' column in the resultset w/ the table alias : 0: jdbc:drill:> select t.`user` from `profiles/2aa32e9e-bdae-8949-8461-c14dafe63ee0.sys.drill` t ; +---------+ | user | +---------+ | apernsteiner | +---------+ Is this expected? I could understand if Drill would only return valid data if I properly aliased and subscripted to get to the element...however: why would it return the logged in username in this query if I neglected to do so? -- Andy Pernsteiner Manager, Field Enablement ph: 206.228.0737 www.mapr.com Now Available - Free Hadoop On-Demand Training <http://www.mapr.com/training?utm_source=Email&utm_medium=Signature&utm_campaign=Free%20available>
