![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
When I did backups last week I had some problems. I started out using a different machine from the usual one -- I was trying to get backups off of my main workstation -- and things were going very strangely, so I shut it down (forcibly, since it seemed to be hung) and moved the drive back to my workstation. There were a couple of anomalies, but things seemed to go OK. This morning (I do backups every Sunday morning) I decided to run a file-system check.
The results were not encouraging. There were a whole bunch of
inodes with bogus block pointers. Weird. Ended up with about 152MB of
debris in lost+found
. Not good. I ran backups for
this week and everything proceeded normally. My best guess is
that the IDE controller on the old machine I was trying to use last week
had problems with large disks. That, or it was just having problems. Or
there might be problems with the drive -- hard to tell.
Bottom line: Dumb bear. If I'd run fsck
last week
before the second attempt, there probably wouldn't have been nearly as
much corruption. This week's backups are probably OK, but many older ones
almost certainly have problems. I need to rewrite my backup scripts to
run fsck
first.