[Pc_Support] Oddball TAR (1) behavior ....

[William A. Mahaffey III]

.... I have several SGI Octanes on my LAN, along w/ 2 Linux PC's & 1 
Win2K Box. 1 of the Linux boxen (2.4 GHz P4, 2 GB RAM, SuSE 9.2, all 
stock) has a largish HDD (160 GB Samsung EIDE 100 HDD, Reiserfs (3.6), 1 
big partition) which I use to back up several of the other machines by 
creating tarballs across my LAN. I have been having problems with that 
process for a couple of weeks, which I *think* I have traced to an 
obscure error (?) message during the process. That message follows:

    .
    .
    .
    .

-rw------- wam/users   1478527 2004-12-28 11:41:42 
test/OUTPUT.MachNoNewLU.2400P4a.gz
tar: test/OUTPUT.MachNoNewLU.2400P4a.gz: File shrank by 479103 bytes; 
padding with zeros

    .
    .
    .
    .

A few dozen entries later, the tar process error-exits with the following:


tar: Error exit delayed from previous errors


The tar process is executing on the Linux box, which has CD (1) 'ed to 
the other box (an SGI Octane, IRIX 6.5.20f) for the backup. The Octane's 
partition is accessed by automount over NFS V3 (both SGI & Linux box) by 
the Linux box. This error seems to cause the Linux box to hang & 
sometimes crash (several times / month). Besides the Rube Goldberg-ish 
nature of this whole process, are there any other known problems w/ 
interactions of tar/NFS/FS-differences which could be causing me 
problems here ? TIA

-- 
	William A. Mahaffey III
---------------------------------------------------------------------
	Remember, ignorance is bliss, but
	willful ignorance is LIBERALISM !!!!