Re: I am sure it is hardware, ufs_inactive: unlinked
On Wed, Jan 15, 2020 at 07:48:04PM -0500, Greg Troxel wrote: > Ronald Georgia writes: > > > I am running 9.99.36 on an old AMD FX-6300 six core processor. It is > > 2009 maybe? Same age on the power supply and memory, but new hard > > drives. It will run fine for a while (couple of days) then completely > > locks up. I had xconsole open and the last time it locked up I had > > this message: > > > > ufs_inactive: unlinked uno on “/“ has non zero size 0 or block … > > Panic: > > > > Questions for all you hardware gurus. > > What would be the most likely suspect? > > In my experience, memory, power supply, and the mains supply if it's not > on a UPS, and the UPS if it is on a UPS. > > > Is there hardware test software that you use and some what trust? > > sysutils/memtestplus. Beware that there is some funkiness with gcc > newer than netbsd5 and multicore tests, that may be resolved but I'm not > sure. > I think the funkiness got worse to the point it doesn't run correctly if built with modern compilers :-( unfortunately I haven't found an easy answer to fixing it. But I suspect netbsd-7 (and possibly netbsd-8?)'s packages work. Other package managers (arch) seem to be downloading the binary & modifying it.
Re: I am sure it is hardware, ufs_inactive: unlinked
Date:Wed, 15 Jan 2020 16:22:12 -0500 From:Ronald Georgia Message-ID: <551ec15b-54a9-4b4e-8788-1450233e5...@gmail.com> | I had xconsole open and the last time it locked up I had this message: It might be hardware, and if the values (if any) in the message (inode number, or anything else) keep varying, then it most likely is, but if you manage to see the same exact message (with numbers, not included in the message, which should be (on one line) something like: ufs_inactive: unlinked ino 1234 on "/" has non zero size 9876 or blocks 456 with allerror 7 where the numbers are all my invention here (and the "/" came from your message, that could also vary possibly) remain constant every time you see the message, then it is probably "just" a corrupted filesystem. In that case, get the system into single user mode, and /sbin/clri /dev/rwd0a 1234 where "/dev/rwd0a" is the raw access device (the 'r') of whatever the "/" filesystem is mounted from (could be sdNx or dkN or ldN or various other things as well as wdNx), check by using the "mount" command (just as "mount" without args, on my system I see /dev/dk6 on / type ffs (log, nodevmtime, local) so I would use "/dev/rdk6" , and the "1234" is the number after "ino" in the message. After the clri command finishes, without doing anything else at all, press the "reset" button (or power cycle the system, if there isn't one of those). Boot into single user mode again, and force a full fsck of the root filesystem fsck -f /dev/rwd0a (or whatever, the same device as the clri) and then reboot.If after this you have no further problems, then all is good.If after a while the problems return, then I would really suspect the hardware. kre
I am sure it is hardware, ufs_inactive: unlinked
All, I am running 9.99.36 on an old AMD FX-6300 six core processor. It is 2009 maybe? Same age on the power supply and memory, but new hard drives. It will run fine for a while (couple of days) then completely locks up. I had xconsole open and the last time it locked up I had this message: ufs_inactive: unlinked uno on “/“ has non zero size 0 or block … Panic: Questions for all you hardware gurus. What would be the most likely suspect? Is there hardware test software that you use and some what trust? -- Ron Georgia “90% of my problems are due to ignorance, the other 10% is because I just don’t know any better.”