Hermes upgrade

Plan to upgrade Hermes by adding 3 new 146gb disks:

See MegaCLI manual at /afs/ugcs/ugcs-admin/mr_sas_sw_ug.pdf

Pre-downtime

 * Create a mail server that only softbounces mail (probably on hephaestus)
 * Add this server as a MX record for our domain
 * Send out downtime announcement
 * vos_dump all mail volumes (just in case)
 * Backup /etc
 * mysql_dump mysql database

Downtime
megacli -PDList -a0 megacli –PDClear -Start -PhysDrv[8:3,8:4,8:5] -a0 megacli -PDMakeGood -PhysDrv[8:3,8:4,8:5] -a0 megacli -LDRecon -Start -R5 -Add -PhysDrv[8:3,8:4,8:5] -ProgDsply -L0 -a0
 * Start softbouncing mail
 * Add disks
 * See if disk exists
 * Plug in and initalize ( See 3.11.6 in manual)
 * Make state "Unconfigured Good"  (See 3.11.4 in manual)
 * Add disks to virtual disk and rebuild (See 3.10.7 in the manual)
 * Re-partition
 * Simply create a new partition on the end
 * Obviously, do not alter existing partitions


 * We will probably have to reboot after this
 * Create new pv on /dev/sda3 and add it to the volume group
 * Set postfix to softbounce and restart it
 * Stop dovecot, webmail, apache, mysql, amavis, spamassassin,
 * Create a new var partition, resierfs (40gb?)
 * Copy old var partition over
 * Restart services
 * Create vicepb partition (leave ~5gb left in VG)
 * Make vicepb filesystem, add it to fileserver

Post-downtime

 * Re-start mail services on hermes
 * Remove softbounce server
 * Test it