Cloud group C - Directory structure maintainence
- Check all the mds.server.id are set correctly for both MDS and replica nodes.
- Check all the mds.inode.start, mds.inode.end and mds.current.inode are set correctly. Note: All the MDS should have unique start and end inode numbers. Also use the same set of inode valued for Replicas of a particular MDS node(optional).
- Check all the mds.replica.info values are set correctly as per the required format in all the machines/nodes.
- Check all the mds.partition.config values are set correctly as per the required format in all the machines/nodes.
- Check all the ips of various nodes are set correct. Note: In the mds..ip key name, the "node id" should concinde with all the mds.server.id provided for all the nodes.
- Check whether all the clients "client.id" are given a unique id.
- Check whether "server.type" in masternode.conf is set to "MDS".
- Check whether "client.ceph.root" is set to the ip of the MDS containing root directory.
Note: Remove/Delete all MDS*.img files in all the node's /data folder to pickup your own copy of generated commands. Run all the MDS and Replica nodes before executing MDS1 node.
- Make sure client.commandsFile file exists in the folder whereever it point to.
- Make sure client.masterIp and client.masterPort are set correct and pointing to the master containing the root folder.
- Make sure the path exists for the server.nfs.folder exist on all the NFS nodes.