>We can't access the backups. Even if we could, we noticed two years ago when trying a restore from the backups, that it doesn't work. Booting the restored server leads to a kernel panic we couldn't figure out. Management said we don't have enough money to fix any of this.
It's often even worse. Many years ago I was responsible for backup a DEC Unix system. Asked for funds to conduct a disaster recover exercise to find out if our backups worked and was turned down even before we had tried to work out how much it would cost.
I often wonder what proportion of backed up files and systems are actually in a state to be restored and have the necessary tools and expertise to do it in a timely manner.
An even worse problem in some places, including the one I worked at, is a tendency for users to assume that backups are kept forever and for the management to totally neglect archiving of important documents such as specifications, drawings, and design calculations of products that have an expected lifetime of over fifty years.
I had to inform several users that I could not restore the file that they had only just noticed was missing because it was last seen more than two years ago and presumably disappeared longer ago than our longest backup cycle which was one year of monthly full backups.
If you can afford 12 months of backups then you can almost certainly afford >=12 years of yearly backups by buying a new set of monthly tapes once a year and taking a set out of rotation and into archive.
(GDPR and earlier laws, etc.)
The best way to delete user data is to encrypt it all with per-user keys so they can be shredded immediately upon request. Backups of user keys themselves can be very short-lived since the keys are static and the lifetime is scoped to GDPR/other laws. Then the encrypted data can be archived indefinitely, so long as encryption keys are never stored in those archives.
Up until now, they didn't know what 'backup not working' meant. Now they know.
Not being bitter -- there was a failure of communication from both parts.
Typically, this kind of decisions are made more or less implicitly during long meetings about priorities, often without any minutes. Managers will presumably try to hold (senior) engineers accountable as they had not made clear enough how important that was.
I obviously don't know who is right here.