Hi all, locally on MacOS I have to set the SPARK_LOCAL_IP environment variable to overcome that very same problem (something along the line of export SPARK_LOCAL_IP="127.0.0.1").
I wonder if it wouldn't be less intrusive to set an environment variable rather than statically setting a SparkConf option in the code as Aron is suggesting. Best regards, Alessandro On Fri, 16 Apr 2021 at 12:26, JiaTao Tao <taojia...@gmail.com> wrote: > Hi > The problem may be the wrong hostname in "/etc/hosts", a way to solve this > problem is to set "spark.driver.bindAddress" explicitly, I've tested this > and it works. > I've created a JIRA(CALCITE-4587) and attached the PR( > https://github.com/apache/calcite/pull/2404/files). > > Regards! > > Aron Tao > > > Vladimir Sitnikov <sitnikov.vladi...@gmail.com> 于2021年4月15日周四 上午2:48写道: > > > > java.net.BindException: Cannot assign requested address: Service > > 'sparkDriver' failed > > > after 16 retries (on a random free port)! Consider explicitly setting > > the appropriate > > > binding address for the service 'sparkDriver' (for example > > spark.driver.bindAddress > > > for SparkDriver) to the correct binding address. > > > > GitHub updates virtual environments from time to time, so the failure > might > > be > > related to a new environment having a different set of network > interfaces. > > > > The failure looks like a true bug rather than a CI glitch. > > > > Vladimir > > >