Which install media for SG-2440?

signal7

Ok - thanks for the note, I'll give that a shot and post my results. I couldn't find any documentation on the appliance itself other than the product specs. Links to download pfSense don't indicate which download fits this hardware.

The 'appliance' nature of this device doesn't make it clear that it would be so fragile in a power outage. Either the root fs should be read only as necessary or the filesystem should be journaled. This problem has been fixed for years, so it surprises me that I would need to buy a ups when one shouldn't have been required. Just my opinion - I bought this for it's reliability. Having to reload after only two months is not a good sign. Please don't take this as negative criticism - it's a suggestion for improvement, if anything.

Also - it's curious that the /etc directory gets hit frequently enough that I found a 3 page discussion of it in the forum with the same error message. If anything, the groups file should be static and unchanging. I know I was sleeping at the time and not even logged into the system.

almabes

NanoBSD is more resilient, but at the cost of limited support of packages.
There's a thread about installing on a ZFS partition instead of UFS, which I am going to look in to.

cmb

The root issue causing that problem is worked around setting sync on /, which we'll be doing by default in 2.2.3 and newer. You can apply it manually if you'd like, by changing the rw part of /etc/fstab to rw,sync so it looks something like this:

# Device                Mountpoint      FStype  Options         Dump    Pass#
/dev/da0s1a             /               ufs     rw,sync              1       1

I'm at upwards of 1000 power cycles on a SG with that set with no problems.

almabes

Just for completeness, this works on other platforms as well, such as the VK-40TE (which I have a few)…correct?

cmb

It works for everything running a full install. nanobsd versions already set sync and so aren't affected.

divsys

Would you say that change is a reasonable retrofit for all non-nano installs 2.x and up or just 2.2.x?

I'm all for improved resilience to power failures…..

stephenw10

Only for 2.2+. Earlier versions built on FreeBSD 8.3 are unaffected.

Steve

divsys

Thanks for that Steve.

I see the underlying cause is discussed in more depth: https://redmine.pfsense.org/issues/4523.

signal7

After futzing with it for a bit, I found what I thought was a much easier way to recover. The last time I did an upgrade, I checked the box for the unit to do a full backup. Ok, so then how do you do a full restore if the web interface doesn't come up? Well, it turns out in my case that if I recovered the groups file from a different system (or from the tgz file in root's home directory) and then disconnected the WAN interface, the GUI would work just fine (very odd, that behavior).

I recovered back to 2.2.1 and then re-upgraded. The system rebooted and was once again unresponsive. Hmmm. It worked fine on 2.2.1 and in theory, the full restore should have overwritten any corrupt files.

I again disconnected the WAN interface and rebooted. The system came up and was responsive on the local interface. I plugged the WAN interface back in. It works fine.

For some reason, if the WAN interface is connected during boot up the system is non functional. I've looked over the boot messages and there are no warnings to speak of.

I may yet do a full reload as suggested, but thought the added data might be interesting to the community - especially if it might resolve some of these problems.

almabes

I had one crap out at a customer location about a month ago. Ended up backing up the config.xml to my laptop, reloading the ADI engineering platform pfSense 2.2.2 and uploading the config. I was done in half an hour. No futzing.

Did the fstab modification last night, and rebooted. All is good.

signal7

At this point I've reloaded it twice. The first time around, I got through getting the config back on there and then added sync to the root fs in the fstab. I rebooted and it went braindead - hung at 'booting from hard disk…' I tried a power cycle and it did the same thing.

Ok - reload it again. This time make the change to fstab, verify it reboots, then do the config. Seems functional at the moment. If I have to reload this again, it's going back to an older release as I'm not really feeling this release is ready.

I try various things and 'futz with it' as a means of learning. Sometimes it really pays off - maybe not today, but the info will come in handy later at some point. You've got to admit, restoring a backup should have done it and should have been much less painless than a complete reload.

almabes

@signal7:

I try various things and 'futz with it' as a means of learning. Sometimes it really pays off - maybe not today, but the info will come in handy later at some point. You've got to admit, restoring a backup should have done it and should have been much less painless than a complete reload.

Me, too.
I was relating a similar tale of woe. Same device. Same problem. I, however, DIDN'T have the luxury of futz time. I had 25 people trying to sell motorcycles, that couldn't because the firewall was down.

I think, when the backup was restored in both our cases, it set the com port configuration back to com1. If you can get to the webConfigurator you can add the tweaks to /boot/loader.conf and get the console back.