Hi,

I'm working with upgrading spark (from 1.6.4 to 2.0.2) and encountered a strange issue when attempting to access the spark ui of a running job:

17/03/28 15:58:24 DEBUG server.Server: REQUEST GET / on HttpChannelOverHttp@750bebc7{r=1,c=false,a=DISPATCHED,uri=/}
...
17/03/28 15:58:24 DEBUG server.Server: RESPONSE /  302 handled=true
...
17/03/28 15:58:24 DEBUG server.Server: REQUEST GET /null/ on HttpChannelOverHttp@750bebc7{r=2,c=false,a=DISPATCHED,uri=/null/}
...
17/03/28 15:58:24 DEBUG server.Server: RESPONSE /null/  302 handled=true
...
17/03/28 15:58:24 DEBUG server.Server: REQUEST GET /null/null/null/ on HttpChannelOverHttp@750bebc7{r=3,c=false,a=DISPATCHED,uri=/null/null/null/}
...
17/03/28 15:58:24 DEBUG server.Server: RESPONSE /null/null/null/  302 handled=true
...
17/03/28 15:58:24 DEBUG server.Server: REQUEST GET /null/null/null/null/null/null/null/ on HttpChannelOverHttp@750bebc7{r=4,c=false,a=DISPATCHED,uri=/null/null/null/null/null/null/null/}

Etc, until a buffer overflow exception.  I'm not familiar with the Spark code base, but I'm slowly digging through the code to attempt to identify the cause. In an application that runs with 1.6.4, the request gets redirected from / to /jobs or so and this does not happen.

Anyone familiar with the code base know where/how the redirect is handled for the request?

Attached is the debug log (from where I think the http request started and ended) if anyone is interested.

--
^___^  Smile!  It's good for you!