Booting from Flash, with a HD in system
-
I'd like to know, is it possible to install the full version onto a CF, USB, or some other flash (Keeping it read-only), and use a Hard Drive for logs and all the other needed uses?
-
Yes but it's not a supported configuration.
You have to edit the fstab manually to mount the drive at boot.
I think the configuration wouldn't last across an upgrade.
I'd be very interested in your experiences setting up logging, squid cache etc.Steve
-
Well, I think setting up Squid, or anything else on the hard drive is a bit beyond my abilities (I know LS, CD .., and that's about it) I guess I'll just have to go with Flash and count on replacing every 6 years or so. I'd never risk a key part of my network on a single, unmonitored hard drive. Well, here's hoping it becomes a supported configuration in the future :)
-
Everyone was 'cd and ls only' at some point. :)
As I remember, I ran out of enthusiasm, it's quite easy to configure squid. It's simply a matter of changing a line in the squid conf file to point at the HD. Of course you have to have the hard drive mounted and partitioned first…and none of this is in the GUI! ::)You should get longer than 6 years from a CF card under 'normal' conditions. Your case fan will probably fail first.
Steve
Edit: Here you go. Tutorial: How to configure squid to use a separate hard drive as its cache .
I seem to remember getting most of the way through that! ;) -
What's wrong with using a hard drive? If you're really worried about it failing, go with a raid 1 setup. And make sure to back up your configurations.
I've been running pfSense(and IPCop before that) with a regular hard drive and never had a problem.
-
Yes I have to agree that if you need to use squid just doing a standard HD install is probably the way to go.
By the time you have installed a separate hard drive you've lost a lot of the advantages of solid state anyway (power consumption, heat, size).Steve
-
Thanks for the help and link. As for the HD, besides personally seeing a number of failures over the years, I do Data Recovery for a living, so my perspective, and opinion of hard drives, is a bit different. Statistically, the single component most likely to fail in nearly any system is the HD. Removing the HD from the equation means I don't have to come home early from a vacation to replace a failed HD in my pfsense box. RAID1 could be an option, but I'm trying to put together a small, cheap, energy efficient box, not another Xeon server. Specifically, I came across a broken eSoft Instagate 404 (You guessed it, internal HD failure) I'd like to put pfsense on. I have a spare 4G IDE flash to put in there, but I hate to waste the capability of the HD. Then again, would pfsense fail if it couldn't access the HD, even if it booted from flash?
-
Then again, would pfsense fail if it couldn't access the HD, even if it booted from flash?
Interesting question, one that I was just considering.
If it failed at boot (likely if you reboot often) then I would think only squid would fail. However if the drive failed at a random time I'm not sure. :-\Personally I've far more PSUs fail than HDs but you're obviously more experienced than me in this matter.
Steve
-
Not to dig too deeply into failure analysis, but I've found a huge difference between crap quality and good quality PS's. The crap capacitors dry out quicker, and the cheap bronze bearings in the fans (Another big failure point) gum up, dry out, and fail. After disassembling about 50 random failed PS's, Capacitors were the biggest failure, and the fans were often on the verge of failure, if not failed. When it fails, a crap PS can also take out the HD, MB, Memory, and even KB and mouse. I've seen it happen more than once.
On the other hand, I wouldn't trust the most reliable drive in the world. Especially with modern drives, it's not a matter of if, but when.