UPGRADE

Upgrade notes for SX 1.0
------------------------

Skylable SX 1.0 contains significant improvements to the storage engine and
is not backward compatible with the beta releases, therefore you need to
follow the upgrade procedure described below. All future stable releases
will either be backward compatible or will provide an automated upgrade tool,
to convert the storage database to the latest version.

We have created a script to help you upgrade from the beta versions (tested
with 0.4 and 0.9). The script will dump the existing cluster structure
(including all user accounts, ACLs, volumes and their contents) and automate
the migration process. We assume the SX was installed using a binary package,
if you installed it differently, you will need to change the example paths.

1. Run the following command:

$ wget -O sxdump.py "http://gitweb.skylable.com/gitweb/?p=experimental.git;a=blob_plain;f=sxdump/sxdump.py;hb=refs/heads/master" && chmod a+rx sxdump.py

It should download the sxdump.py script. You can edit it and change the
BACKUPDIR (it needs to provide enough space to fit the cluster's data).

Now run './sxdump.py sx://[admin@]cluster' and it should create two scripts:
old.sh and new.sh

2. Run 'old.sh' and make sure there were no errors when backing up the data.
IMPORTANT: Encrypted volumes will not be backed up, they have to be handled
manually.

3. On all nodes run 'sxserver stop' and move the previous storage directory
into another location, eg.
# mv /var/lib/sxserver/storage /var/lib/sxserver/storage.old
(Since you have already prepared a backup, you can also remove it if you're
short on space.)

Now install the SX 1.0 release, using binary packages or from the source code.

4. Following the order the nodes were added to the cluster for the first
time, run this command on each node:

# echo SX_CFG_VERSION="2" >> /etc/sxserver/sxsetup.conf
# sxsetup --force-reinit --config-file /etc/sxserver/sxsetup.conf

5. The new cluster should be up and running! It will have the same admin
account as the original cluster. Now run the 'new.sh' script and wait
until it recreates all users, volumes, permissions and re-uploads the data.

Please let us know if you have any problems or questions.