[mythtv-users] Mythtv FE in a frequent power loss environment

[Ronald Frazier]

> Err, the statement above is correct - a client crash should not be able to
> damage the **file system**.

Yes, thanks for actually catching what I was saying. :-)


> Alternatively, put all the logging on it's own filesystem and be
> prepared to fix it (ie reformat it) if it gets damaged.
>
> However, I've found ext3 to be very robust, and ext2 isn't bad. I rather
> think that if you use ext3 then you're unlikely to have any problems.

But at this point, if you need to do this, adding the hard drive to
the frontend is going to be pointless. As we just said, an NFS mount
may get corrupt files, but not a corrupt filesystem (since the
assumption is the server will have a UPS). A local hard disk will
likely get the same corruption in addition to a corrupt filesystem. So
this is going to be no better than an NFS mount in this respect
(worse, actually).

Plus, an NFS mount is easier to "reimage". Just login to the server,
make a backup of the frontend's root folder (tar/gzip it up), and if
it gets corrupt, just rm -r the folder and then replace it from the
tgz file. I do this all the time when I'm experimenting with my
diskless frontends. Make a copy, apply some changes, and if I break
anything just revert back to the original. Actually, I duplicate the
folder so I have a production folder and a dev folder, and the folder
that gets mounted on the frontend is just a symlink to whichever one I
want active. This was I can do testing over the course of several days
switching it back and forth between prod and dev whenever my wife
needs to use it.


-- 
Ron
Ronald Frazier Photography - http://www.ronfphoto.com/
Blogging About Photography - http://ronfrazier.blogspot.com/