I have had this problem on 2 servers and I want to know how should I handle the problem. My server has an SSD (/dev/sda) and two SATA drives (/dev/sdb and /dev/sdc).
The second drive (/dev/sdb) has problem and sometimes the drive is there but filesystem has problem and sometimes the device totally disappears.
Case 1- When the drive is there (ls /dev shows it) but filesystem has problem and I reboot the server (because device is busy and cannot be fsck-ed.), the server will not boot because some services have dependencies on that drive and those services somehow halt boot process.
Drives are mounted in /etc/fstab onto /sdb1 and /sdc1 directories.
Case 2- When the drive (/dev/sdb) disappears,the third drive (previously /dev/sdc) is renamed to /dev/sdb and you can guess it causes problems because that is not the drive my programs expect to see. Again the server will not boot because programs do not see their data on the new sdb.
Question 1: How may I configure the server, so that it boots even if some services cannot see their dependencies on failed drives?
Question 2: How can I avoid drives from appearing with different /dev/name if other drives disappear (failed hardware etc.)?
smartctl --all /dev/sdb
in your question to be 100% sure what you're facing is imminent drive failure? (now only 99% certain).