On Dec 11, 2009, at 6:13 AM, Jim Pingle wrote:
On 12/11/2009 10:50 AM, David Burgess wrote:
I've been happily using 1.2.3-RC1 for many months now on a Soekris
net5501 and a 100GB 2.5" SATA drive. I like the idea of an embedded
system on a CF card, but that's not possible or advisable for me as
I'm running the squid and freeswitch packages.
I was wondering however, if it would be difficult, inadvisable, or of
no advantage to hack together an embedded system to run from a
read-only CF card that mounts certain filesystems on writable media,
such as a hard drive, where temp data such as disk cache and audio
recordings would live.
I've thought a bit about this in the past, and it might be doable in
the
future or via some kind of filesystem management package, if someone
were to come up with one, but it isn't something that would be
recommended (at least not yet) or supported.
I don't know a tonne about the innards of pfsense and I've never
played with the nanoBSD version. Is this something that would work in
principle? Would it exploit the benefits of a read-only root
filesystem
(cold-reset resiliency,
The moment you have a drive mounted rw, you lose this. :-)
improved fs security, system
responsiveness)? Would it require a lot of messing, besides manually
altering /etc/fstab?
You'd also have to alter the packages (or create appropriate
symlinks if
they can be followed by the application) to point those directories or
files at the new storage location. Some packages might have built-in
path settings and you'd just need to change the paths and hit save.
Otherwise, you may need to alter the code for the package.
As with most things, if you want to experiment, it's up to you, but do
so with caution (and plenty of backups) and remember that you'll be
out
on a limb without a net to catch you if something breaks.
Jim
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
Commercial support available - https://portal.pfsense.org
I might be missing the boat here, but what about using a 2.5" SSD
instead of flash + normal HD? That way you get the benefit of solid
state, plus you have the space & performance for a regular file system
so you can run all the packages you want. Granted, SSDs aren't the
cheapest things around, but it seems like a simpler solution.
I've been considering an SSD paired with a 19" Supermicro case + intel
atom that was pointed out in another discussion thread. Besides the
cost of the SSD, can anyone fill me in on why an SSD wouldn't be good
for running the full version of PFsense with packages?
Jeremy
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
Commercial support available - https://portal.pfsense.org
