[ https://issues.apache.org/jira/browse/HDDS-4107?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17176745#comment-17176745 ]
Glen Geng edited comment on HDDS-4107 at 8/13/20, 3:48 AM: ----------------------------------------------------------- Thanks for looking at this Jira [~msingh], [~elek] [~timmylicheng] [~nanda] Please also have a look at this issue. For the non-SCM-HA mode, if want to migrate from version 0.6.0 to the version that contains this change, OM/SCM is not affected, we need do something at Datanode side. After shutdown Datanode process and replace the manifest, need an upgrading script to rename the directory from "hdds/scmUuid" to "hdds/clusterID". The script can decide what needs to do according to the scmUuid and clusterID recorded in version file. For HA-mode, customers need a boot-up process, we can do above things during the boot-up process to enable the HA mode. was (Author: glengeng): Thanks for looking at this Jira [~msingh], [~elek] [~timmylicheng] [~nanda] Please also have a look at this issue. For the non-SCM-HA mode, if want to migrate from version 0.6.0 to the version that contains this change, OM/SCM is not affected, we need do something at Datanode side. After shutdown Datanode process and replace the manifest, need an upgrading script to rename the directory from "hdds/scmUuid" to "hdds/clusterID". The script can decide what needs to do according to the scmUuid and clusterID recorded in version file. For HA-mode, customers need a boot-up process, we can do above things to enable the HA mode. > replace scmID with clusterID for container and volume at Datanode side > ---------------------------------------------------------------------- > > Key: HDDS-4107 > URL: https://issues.apache.org/jira/browse/HDDS-4107 > Project: Hadoop Distributed Data Store > Issue Type: Sub-task > Components: SCM > Reporter: Glen Geng > Assignee: Glen Geng > Priority: Major > Labels: pull-request-available > > The disk layout per volume is as follows: > {code:java} > ../hdds/VERSION > ../hdds/<<scmUuid>>/current/<<containerDir>>/<<containerID>>/metadata > ../hdds/<<scmUuid>>/current/<<containerDir>>/<<containerID>>/<<dataDir>>{code} > However, after SCM-HA is enabled, a typical SCM group will consists of 3 > SCMs, each of the SCMs has its own scmUuid, meanwhile share the same > clusterID. > Since federation is not supported yet, only one cluster is supported now, > this Jira will change scmID to clusterID for container and volume at Datanode > side. > The disk layout after the change will be as follows: > {code:java} > ../hdds/VERSION > ../hdds/<<clusterID>>/current/<<containerDir>>/<<containerID>>/metadata > ../hdds/<<clusterID>>/current/<<containerDir>>/<<containerID>>/<<dataDir>>{code} -- This message was sent by Atlassian Jira (v8.3.4#803005) --------------------------------------------------------------------- To unsubscribe, e-mail: ozone-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: ozone-issues-h...@hadoop.apache.org