NFS mounts fail whilst amanda is running

[barry]

I've got a bit of an odd issue backing up an RHEL6 client (amanda-2.6.1p2) to an Scientific Linux 5 Server (amanda-2.5.0p2).

We have a variety of cron jobs that call scripts from automounted NFS shares, whilst amanda is running the mounting of these shares fail (even manually)!

The problem seems to persist even when I switched to using a test server running SL6 (amanda-2.6.1p2) and also whilst running a downgraded client.

An strace of the runtar process reveals an awful lot of error messages similar to;

bind(7, {sa_family=AF_INET, sin_port=htons(891), sin_addr=inet_addr("0.0.0.0")}, 16) = -1 EADDRINUSE (Address already in use)

The mount command returns an Input/Output error.

Can anyone think of what might be causing this?

[barry]

"An strace of the runtar"

should have read

"An strace of the tar command during sendbackup"

[barry]

I can reproduce this error on demand now by enabling/disabling nis using;

authconfig --update --enablenis
authconfig --update --disablenis

the backup completes in about 1/4 of the time when NIS is disabled.

[barry]

This is being caused by NIS group lookups, changing the group line in nsswitch.conf from;

group: file nis

to

group: file

stops the errors from occurring.

How to fix this is a different matter..