>... I think it times out because there is over
>100,000 files and it can't create the file list. ...
Does the whole file system has 100,000 files, or just /raid1/production?
Using GNU tar, it would only matter what's under /raid1/production.
>The "production"
>directory isn't even that big, it's only around 8 GB.
Size doesn't matter for estimates. A stat() of a 10 KByte file takes
the same amount of time as a 10 GByte file.
>I increased the timeout period in the amanda.conf file which I thought
>would help, but it didn't. ...
Which timeout variable did you adjust? It should have been "etimeout".
>When I manually ran amdump, the gtar process
>list took up like 60 to 70 percent of the CPU usage. I took a look 20
>minutes later and it was still running but the CPU usage was only 0.4
>percent so I just killed it. I don't think it would've ever finished.
I don't understand. If the CPU usage dropped that much, it implies the
GNU tar might have been done. Did you do a "ps" to see what was going on?
You might also take a look at /tmp/amanda/sendsize*debug.
>Before amanda, I used dump to do level zero's every night on the
>/raid1/production directory which worked well. Can I do that with
>amanda, I don't mind getting a level 0 backup everyday.
I don't think you can do this. Amanda will see that you are using "dump"
and convert the mount point to a disk name, then pass that to "dump"
which will do the whole file system.
You could probably do the infamous GNU tar wrapper hack from:
ftp://gandalf.cc.purdue.edu/pub/amanda/gtar-wrapper*
only in your case you'd detect when the request was for /raid1/production
and run dump instead. You might also set this up with its own dumptype
and configure it with "dumpcycle 0" so Amanda understands it is going
to do a full dump every time.
>Ajay
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
