The stderr should be in the same directory as the rest of the Hadoop userlogs (e.g., syslog). If your Hadoop files are in "~/hadoop_data/hadoop_local-username/", then the metrics are outputted to "~/hadoop_data/hadoop_local-username/userlogs/job_yyyymmddhhmm_nnnn/*/stderr" by default. Also, if you're in a distributed setting, each machine will have its own stderr log.
Young On Fri, Oct 10, 2014 at 4:26 PM, xuhong zhang <xuhongne...@gmail.com> wrote: > I can not see the output from my command line, does stderr here really > mean the terminal output? > > sorry, i really don't know where to locate the output? > > thanks > > On Fri, Oct 10, 2014 at 2:16 PM, Young Han <young....@uwaterloo.ca> wrote: > >> Use them as -Dgiraph.metrics.enable=true, after GiraphRunner but before >> you specify the algorithm of interest. In other words, >> >> hadoop jar .... org.apache.giraph.GiraphRunner \ >> -Dgiraph.metrics.enable=true \ >> -Dgiraph.metrics.directory=dir \ >> org.apache.giraph.examples.SomeAlgorithm \ >> -ca AlgorithmSpecificOpts -vif InputFormat ... etc. >> >> As Matthew mentioned, by default it outputs metrics to the stderr >> logs---I haven't used the directory option before either. >> >> Young >> >> On Fri, Oct 10, 2014 at 2:10 PM, Matthew Saltz <sal...@gmail.com> wrote: >> >>> Use the -ca flag, for each one. Metrics.enable puts metrics from the run >>> in the stderr file of each worker by default. I've never used the command >>> to change the directory. If that's not what you're looking for let me know, >>> I'll be glad to help. >>> >>> Best, >>> Matthew >>> El 10/10/2014 19:55, "xuhong zhang" <xuhongne...@gmail.com> escribió: >>> >>>> Dear everyone, >>>> >>>> How to use "giraph.metrics.enable=true" and >>>> "giraph.metrics.directory=dir" options. >>>> >>>> Where should i specify them in the command? >>>> >>>> thanks >>>> >>>> -- >>>> Xuhong Zhang >>>> >>> >> > > > -- > Xuhong Zhang >
