Update: One of the 3TB Seagate Storage Drives was dying and any time I had it connected to the system, I got the IO/read errors, eventually even when the system was booting.
As described below, booting to an old kernel gave me several hours of reliable service so that I could do a backup of the most recent data on drive, and now it is inaccessible by any system I hook it into. It's backed up, so no worries there.
Nothing about the scenario makes a lot of sense, because removing a storage drive (not the system disk) seems to have fixed all issues. Why would a dying drive behave this way? I think it was dumb luck that booting into older kernel gave me good results.
Can anyone explain how this storage drive failure caused all of these problems?
Have I inadvertently solved a different problem?
We have an old Dell Optiplex running Ubuntu 14.04 LTS as a file and ftp server for the company. I had done updates and it had appeared to be running without problems on newly installed software from apt and new kernels, etc. The machine wasn't even rebooted since last night, but this morning it's had this host of problems:
- Connecting over SSH works but
ls
gave me an IO error before reboot.- Rebooting/shutdown now makes the machine get stuck. Error is
rpcbind: rpcbind terminating on signal.
- After a hard reboot, usb keyboard seems to be sending incorrect information to PC, as my password doesn't appear to work when I log in, but SSH works as expected, with or without SSH keys.
- In newer versions of kernels, I cannot
less
most files, although after a rebootls
works and I can open all directories without a lot of trouble.- In newer kernels, SMB and Netatalk can connect but the machines give generic "permissions" or other random errors to say they can't open files that they can see.
- I have booted to
3.8.0-29-generic
and these problems seem resolved, but this is like putting on a spare tire.I have run the hardware diagnostic with no warnings, so I don't think there are problems with the PC hardware, mobo, SATA cables, HDD, etc. It appears to be a software problem since the older kernel seems to "fix" the issue, but I'm not ruling out intermittent hardware failure.