[jira] [Commented] (FLINK-9848) When Parallelism more than available task slots Flink stays idle
[ https://issues.apache.org/jira/browse/FLINK-9848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16804860#comment-16804860 ] Till Rohrmann commented on FLINK-9848: -- Is this still an issue [~yazdanjs]? If not, then please close this JIRA issue. > When Parallelism more than available task slots Flink stays idle > > > Key: FLINK-9848 > URL: https://issues.apache.org/jira/browse/FLINK-9848 > Project: Flink > Issue Type: Bug > Components: Runtime / Coordination >Affects Versions: 1.5.0, 1.5.1 >Reporter: Yazdan Shirvany >Assignee: Yazdan Shirvany >Priority: Critical > > For version 1.4.x when select Parallelism > Available task Slots, Flink throw > bellow error right away > {{NoResourceAvailableException: Not enough free slots available to run the > job}} > > but for version 1.5.x there are two different behaviors: Sometimes job ran > successfully and sometimes it throw bellow error after 5 minutes > > {{org.apache.flink.runtime.jobmanager.scheduler.NoResourceAvailableException: > Could not allocate all requires slots within timeout of 30 ms. Slots > required: 5, slots allocated: 2}} -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (FLINK-9848) When Parallelism more than available task slots Flink stays idle
[ https://issues.apache.org/jira/browse/FLINK-9848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16547822#comment-16547822 ] Till Rohrmann commented on FLINK-9848: -- I think this is the correct behaviour. When running a batch job, you don't always have to deploy all tasks in parallel. Sometimes it is also possible to execute parts of the job sequentially while keeping the intermediate results around. > When Parallelism more than available task slots Flink stays idle > > > Key: FLINK-9848 > URL: https://issues.apache.org/jira/browse/FLINK-9848 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0, 1.5.1 >Reporter: Yazdan Shirvany >Assignee: Yazdan Shirvany >Priority: Critical > > For version 1.4.x when select Parallelism > Available task Slots, Flink throw > bellow error right away > {{NoResourceAvailableException: Not enough free slots available to run the > job}} > > but for version 1.5.x there are two different behaviors: Sometimes job ran > successfully and sometimes it throw bellow error after 5 minutes > > {{org.apache.flink.runtime.jobmanager.scheduler.NoResourceAvailableException: > Could not allocate all requires slots within timeout of 30 ms. Slots > required: 5, slots allocated: 2}} -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (FLINK-9848) When Parallelism more than available task slots Flink stays idle
[ https://issues.apache.org/jira/browse/FLINK-9848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16543681#comment-16543681 ] Yazdan Shirvany commented on FLINK-9848: Actually I re-tried with multiple examples and seems only {{batch/WordCount.jar}} can be successful when {{parallelism > available slots}} > When Parallelism more than available task slots Flink stays idle > > > Key: FLINK-9848 > URL: https://issues.apache.org/jira/browse/FLINK-9848 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0, 1.5.1 >Reporter: Yazdan Shirvany >Assignee: Yazdan Shirvany >Priority: Critical > > For version 1.4.x when select Parallelism > Available task Slots, Flink throw > bellow error right away > {{NoResourceAvailableException: Not enough free slots available to run the > job}} > > but for version 1.5.x there are two different behaviors: Sometimes job ran > successfully and sometimes it throw bellow error after 5 minutes > > {{org.apache.flink.runtime.jobmanager.scheduler.NoResourceAvailableException: > Could not allocate all requires slots within timeout of 30 ms. Slots > required: 5, slots allocated: 2}} -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (FLINK-9848) When Parallelism more than available task slots Flink stays idle
[ https://issues.apache.org/jira/browse/FLINK-9848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16543395#comment-16543395 ] Till Rohrmann commented on FLINK-9848: -- Under which conditions (apart from {{parallelism > available slots}}) does the job run successfully (assuming that you cannot start new {{TaskManagers}} dynamically like on Yarn or Mesos)? > When Parallelism more than available task slots Flink stays idle > > > Key: FLINK-9848 > URL: https://issues.apache.org/jira/browse/FLINK-9848 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0, 1.5.1 >Reporter: Yazdan Shirvany >Assignee: Yazdan Shirvany >Priority: Critical > > For version 1.4.x when select Parallelism > Available task Slots, Flink throw > bellow error right away > {{NoResourceAvailableException: Not enough free slots available to run the > job}} > > but for version 1.5.x there are two different behaviors: Sometimes job ran > successfully and sometimes it throw bellow error after 5 minutes > > {{org.apache.flink.runtime.jobmanager.scheduler.NoResourceAvailableException: > Could not allocate all requires slots within timeout of 30 ms. Slots > required: 5, slots allocated: 2}} -- This message was sent by Atlassian JIRA (v7.6.3#76005)