@ Fajar
yes exactly that would be a lot faster , and ya as Vidal has mentioned, this whole process goes very deep when it comes to complex applications like databases and servers , so going for a superficial log file based rollback is a bad idea, i should be thinking in ur lines now , ......
On 4/2/08, Fajar A. Nugraha <fajar@xxxxxxxxx> wrote:
Satish Eerpini wrote:
and ya Fajar, i actually was reluctant to use yum in my first implementations of the scripts , but didn't think of a better method given my low knowledge, ..... actually using storage and providing backup and recovery was a option i already had, .... but my point was to design it to be very simple to use and a very 'swift' application ........
To comment about "swift",
that's part of the reason why I mentioned lots of stuff about zfs in my earlier post.
Take for example, the simplest scenario :
- you have opensolaris (SXCE) dom0
- you have Linux domU, using zfs-emulated-volume-backed storage, located on (for example) zpool/linux-root
A more complex setup could involve iscsi, but I won't cover that now.
A "backup" would be as simple as executing the command
"zfs snapshot zpool/linux-root@backup"
which would be completed almost immediately
and a "rollback" would be as simple as :
- shutting down domU
- zfs rollback zpool/linux-root@backup
- starting up domU
which would only take roughly the time required to restart a domU.
I believe that's lotta-quicker, simpler, and more reliable compared to any rpm-based rollback scenario.
Regards,
Fajar
_______________________________________________
Yum mailing list
Yum@xxxxxxxxxxxxxxxxxxxx
https://lists.dulug.duke.edu/mailman/listinfo/yum
_______________________________________________ Yum mailing list Yum@xxxxxxxxxxxxxxxxxxxx https://lists.dulug.duke.edu/mailman/listinfo/yum
