Multiple SG-5100 OS corrupted after power drop
-
We deployed a Netgate SG-5100 with 2.4.5 to a client that has some power issues - semi-frequent power outages lasting longer than the UPS does. Every few times after the power comes back on the Netgate loses its OS and comes up to the boot: prompt on the console. We have swapped it out with another of the SG-5100's running 2.4.5 and experience the same thing. Is there some config that can remedy this or should I get replacements for all 5 of the units I am getting ready to deploy?
-
I have been trying to replicate it back here at my office with the original unit and most times it reboots fine, sometimes I get a boot loop, other times I get the db> prompt.
-
@cmlong89 this isn’t a defect with the 5100. These are not your typical home routers and are sensitive to having the power yanked from them suddenly, just like a PC would be. You need to invest in a better/bigger UPS unit or dedicate one just for the router itself if you are also using the UPS for switches or other devices. That should get you longer run time obviously depending on the UPS battery capacity. You could also run the NUT package which can be configured to monitor the UPS battery level and then tell the router to make a clean shutdown if needed.
-
You could wipe and reload them using ZFS as the filesystem instead of UFS. ZFS is more resilient when it comes to those types of failures.
-
@jimp my understanding was that zfs wasn’t recommended for the onboard storage. Is that not the case after all? I would have rolled over to zfs ages ago myself but didn’t see the need to install a separate m.2. Why mess with something that works, you know?
-
There may be more writes involved with ZFS but generally speaking on the current systems we ship that shouldn't be a problem even on the onboard storage. Lots of us run every device we can using ZFS without issues.
Eventually, ZFS will be the default for factory installations as well.
-
@jimp that’s great to know. I myself prefer zfs but had been operating under the belief that it was less optimal for using with the onboard storage. Guess I have myself a 10 minute upgrade to conduct today. Thanks for the information!
-
-
@cmlong89 just updated my system to zfs a short while ago. Very glad @jimp clarified that matter for me/us. Cannot overemphasize the value of having a good UPS that is adequate for your power needs. The number of people I know who would use a small UPS (meant only to allow you to quickly shut down a computer in the event of an outage) for multiple devices and then complain about bad run time or hard shutdowns always has perplexed me.
