Hi Yuchen, NOT_FOUND means predicator was failure. I guess you configure "*predicates*" when creating a measure. So Griffin will check whether predicate is successful before submitting a job. If it fails with several attempts (default repeat is 12 times), Griffin will set the instance status to NOT_FOUND.
Thanks, Kevin Yao On Wed, Apr 24, 2019 at 4:35 PM Yuchen Zhang <[email protected]> wrote: > Hi William, > > Sure, I just attached the screenshot for the NOT_FOUND state I get from > the web UI. I have no idea about the meaning of the state, but I guess the > "NOT_FOUND" should mean that the service can't find the result in Hadoop. > Since there's no application in Yarn, I think the problem should be in the > job submission. > > Thanks, > Yuchen Zhang > > -----Original Message----- > From: William Guo <[email protected]> > Sent: Tuesday, April 23, 2019 2:51 PM > To: [email protected] > Subject: Re: problems on griffin docker > > hi Yuchen, > > We didn't see your log or screenshot, could you attach them? > > Thanks, > William > > On Mon, Apr 22, 2019 at 9:27 PM Yuchen Zhang > <[email protected]> > wrote: > > > Hi there, > > > > > > > > I used the docker image provided in the Griffin doc to build a service. > > Everything goes well until I submit a test job to this service. It’s > > expected to get a result after some minutes but I got NOT_FOUND state > > for this job. I’ve checked the resource manager in Hadoop and found > > there’s no application, so I think there should be some problem for > > the service to submit job to spark through Livy. > > > > > > > > Any ideas about it? > > > > > > > > Thanks, > > > > Yuchen Zhang > > > > > > >
