1. Introduction

• Volume databases (text)
• /vice/vol/VolumeList Holds partition and volume information for volumes for which this server is custodian. This is file is generated by srv.
• /vice/vol/groupid is maximum replicated volume id allocated so far.
• HOSTS (text)
  • Entries describe volume storage groups: id hostname1 hostname2 etc. Stored in /vice/db/VSGDB
  • /vice/db/hosts servers should be in this file. Servers found in the VSGDB are checked to be in this file before proceeding to make a volume.
  • /vice/db/servers contains an id for each server.

Createvol_rep works as follows: Parameters are: <volname> <vsgaddr> <partitionname> [ <groupid> ]

• check we are the SCM
• Check that the volname doesn't appear in AllVolumes or VRList
• Get the list of servers from the vsgaddr using the VSGDB
• verify the servers against /vice/db/hosts
• get a new replicated id for this replicated volume using the /vice/vol/groupid file
• use volutil to create the volume on each host, using the partition, volumename and groupid acquired before. Note that each server now assigns a volume id for this newly created volume. It dumps

1.1 bldvldb.sh

• get /vice/vol/VolumeList from each server in /vice/db/hosts. Names is server-name.list.new. First try this using ftp, then rfs, then krcp.
• if the file was acquired move it to /vice/vol/remote/server-name.list
• dump them all in /vice/vol/BigVolumeList
• volutil rebuilds the VLDB with volutil makevldb /vice/vol/BigVolumeList

Note this rebuilds the VLDB on the machine where is script is run, i.e. the scm.