/proc is not an actual file or directory it is just a place that holds info on what is currently running

/proc and /sys on RHEL are extensive directory trees of files which can be read (about 7k entries under /proc).  They are all size 0 but when you cat them there's info there from the kernel.  I would have thought Networker would know by default not to try to back them up but it was.  One of my LTO-5 tapes it was putting /proc on showed a capacity of 31TB used.  Something weird was happening.

So I added to the Unix Standard Directives skip proc and skip sys.

the savefs probe would the one to verify with.   It should automatically exclude.    The expansion/client config wizard show everything regardless of whether it will be backed up or not.

savefs/save will automatically exclude /proc and /sys when saveset ALL is specified in the client resource.    If you've explicitly referenced /proc and /sys in the savset then they will be backed up.

The first time I set a client up I didn't use ALL and this was when I noticed /proc was being backed up.

But even if you use ALL when you expand the / directory in the client properties (or via the configure wizard) you see that /proc is still checked.

Hi Avhays,

The answer is very simple. Just make a skip directive (given below) and apply it in client properties' directive field (at General tab).

<>

skip: *.*

<>

skip: *.*

...and here u go.

How do I run the savefs probe?  If I run 'savefs -p' on the client it just prints out the filesystems, not individual files.

you can run savegrp with option -p from the server for the client in question:

savegrp -vvvp -c -G

please ensure that the saveset ALL is specified for the client

With save set All, no data under /proc and /sys will be captured.  You can verify that with mminfo.  Both filesystems are specific (proc and sysfs) and as such are not part of regular backups.