[jira] [Commented] (MESOS-3045) Maintenance information is not populated in case of failover
[ https://issues.apache.org/jira/browse/MESOS-3045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14680946#comment-14680946 ] Joseph Wu commented on MESOS-3045: -- Tentative review: https://reviews.apache.org/r/37314/ > Maintenance information is not populated in case of failover > > > Key: MESOS-3045 > URL: https://issues.apache.org/jira/browse/MESOS-3045 > Project: Mesos > Issue Type: Task > Components: master >Reporter: Joseph Wu >Assignee: Joseph Wu > Labels: mesosphere > > After a master has failed, it must re-populate maintenance information (i.e. > from the registry to the local state). > Some methods In src/master/master.cpp: > * Master::recover > * Master::reregisterSlave > Possible test(s): > * src/tests/master_tests.cpp::RecoverResources - Add an extra check for > filling in the maintenance information. > * src/tests/master_tests.cpp::RecoveredSlaveReregisters - Add an extra check > for filling in the maintenance information. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (MESOS-3045) Maintenance information is not populated in case of failover
[ https://issues.apache.org/jira/browse/MESOS-3045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14680629#comment-14680629 ] Joseph Wu commented on MESOS-3045: -- No. Agents are mostly agnostic to maintenance. (Framework schedulers live on the master, and *are* made aware of maintenance.) This issue tracks the transfer of information from the persistent storage (registry) to the master's local state. > Maintenance information is not populated in case of failover > > > Key: MESOS-3045 > URL: https://issues.apache.org/jira/browse/MESOS-3045 > Project: Mesos > Issue Type: Task > Components: master >Reporter: Joseph Wu >Assignee: Joseph Wu > Labels: mesosphere > > After a master has failed, it must re-populate maintenance information (i.e. > from the registry to the local state). > Some methods In src/master/master.cpp: > * Master::recover > * Master::reregisterSlave > Possible test(s): > * src/tests/master_tests.cpp::RecoverResources - Add an extra check for > filling in the maintenance information. > * src/tests/master_tests.cpp::RecoveredSlaveReregisters - Add an extra check > for filling in the maintenance information. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (MESOS-3045) Maintenance information is not populated in case of failover
[ https://issues.apache.org/jira/browse/MESOS-3045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14679516#comment-14679516 ] Klaus Ma commented on MESOS-3045: - [~kaysoky], regarding Master::reregisterSlave, are we storing the maintenance information in each slave and re-send to master when re-register? > Maintenance information is not populated in case of failover > > > Key: MESOS-3045 > URL: https://issues.apache.org/jira/browse/MESOS-3045 > Project: Mesos > Issue Type: Task > Components: master >Reporter: Joseph Wu > Labels: mesosphere > > After a master has failed, it must re-populate maintenance information (i.e. > from the registry to the local state). > Some methods In src/master/master.cpp: > * Master::recover > * Master::reregisterSlave > Possible test(s): > * src/tests/master_tests.cpp::RecoverResources - Add an extra check for > filling in the maintenance information. > * src/tests/master_tests.cpp::RecoveredSlaveReregisters - Add an extra check > for filling in the maintenance information. -- This message was sent by Atlassian JIRA (v6.3.4#6332)