Move a VM from one cluster to another

• set dnsTTL to 30 in ldap for adayevskaya
• on the target create a new LV to accept the instance lvcreate -n install-adayevskaya -L 5g vg_ganeti
• on the source, shut down the instance gnt-instance shutdown adayevskaya.debian.org·
• give the source host ssh access to the target:
  • cat /etc/ssh/ssh_host_rsa_key.pub
  • tee -a /etc/ssh/userkeys/root
  • iptables -I INPUT -s bm-bl1 -j ACCEPT
  • ip6tables -I INPUT -s bm-bl1 -j ACCEPT
• copy it over: cat /dev/mapper/adayevskaya | ssh root@manda-node03 'cat > /dev/vg_ganeti/install-adayevskaya' # pidof cat, and cat /proc/<pid>/fdinfo/[01] to check on progress
• clean up ssh and the firewall again: puppet agent -t ferm -i /etc/ferm/ferm.conf
• resize partitions if the new block device is larger using fdisk. Delete and create the partition at the same starting position.
• open the partition table: kpartx -a -v -p -part /dev/vg_ganeti/install-adayevskaya
• fsck the filesystem if required, resize if necessary fsck.ext4 /dev/mapper/vg_ganeti-install--adayevskaya-part1 resize2fs /dev/mapper/vg_ganeti-install--adayevskaya-part1
• mount the partition and fix etc/network/interfaces: mkdir /mnt/target-adayevskaya mount /dev/mapper/vg_ganeti-install--adayevskaya-part1 /mnt/target-adayevskaya cd /mnt/target-adayevskaya vi etc/hosts etc/network/interfaces cd / umount /mnt/target-adayevskaya rmdir /mnt/target-adayevskaya
• remove partitions kpartx -d -v -p -part /dev/vg_ganeti/install-adayevskaya
• create the instance, adopting the block device gnt-instance add \ --no-ip-check \ --no-name-check \ -t blockdev \ --no-start \ -t plain \ --disk=0:adopt=install-adayevskaya \ --backend-parameters memory=2g,vcpus=2 \ --os-type noop \ -n manda-node03 \ adayevskaya.debian.org gnt-instance modify \ --disk-template drbd \ -n manda-node04.debian.org \ adayevskaya.debian.org
• start the instance, check everything works gnt-instance start adayevskaya.debian.org gnt-instance console adayevskaya.debian.org
• update ud-ldap (ip addresses, physicalHost, sponsor, location, dnsTTL) and ud-generate
• ud-replicate on denis
• check puppet and ssh authkeys on all hosts for the old IP address
• update nagios
• update dsa-passwords
• update reverse DNS
• zero volumes on the old cluster
• remove the instance on the old cluster
• remove volumes from puppet/multipath/storage arrays and our google docs LUN map