Amanda Users: I have been really lucky so far with using Amanda for the past 3 years without any problems, but I recently ran into a stumper that I cannot figure out. I just built a new Backup server on a RedHat 7.1 machine in which I back up a mixed environment of Linux, Tru64 UNIX, Solaris, IRIX, and now thanks to Samba, Windows 2000. I built the server from the source on the CVS tree last week (9/20/01). I am only having a problem with 2 of my clients out of about 60. These 2 clients just happen to be running RedHat 6.1 with a 2.2.16 kernel. I used to back these up without problems with the old server which was running Redhat 6.0 and amanda 2.4.2 from the CVS tree back in September 1999(interestingly, exactly 2 years ago to the day of my latest CVS snapshot).
Now for the problem: Each of these clients has more than one partition that gets backed up as a seperate entry in the disklist file. Some of the partitions do get backed up, but some seem to just stop in the middle of the backup. The Server will wait all day for the clients because it seems that the server can still communicate with the clients. On the client machines the sendbackup/gzip process goes from using 93% of the CPU to 0% of the CPU, but it still is running (according to top). It is not getting stuck at any particular place on the disk because I can split the disks up even smaller and I still get the problem. Also I have ran the command from runtar to /dev/null and completes without problem. The machine does not run out of memory or swap space and the sizes of the partitions that will not backup range from 200mb to 2gb. Does anyone have any ideas as to what might be going wrong? The extent of my debugging capabilities is through the use of the log files because I have not had to troubleshoot much in the past. Thank You, ~Sam Salvatore J. Guercio UNIX Systems Administrator Internal Information Systems - Platform Engineering Northrop Grumman Corporation - Electronic Systems (716) 631-0610 Ext. 331 [EMAIL PROTECTED]
