Problem while taking backup of Mounted filesystems

[Rajesh R_2]

Hai guys,

I have a problem. I have two servers one HP-UX and one
Linux Server. I take tape backup of these two servers
daily. My tape drive is on my HP-UX server. I mount the
Linux server filesystems on my HP-UX server using
automount feature and take backup daily. But now the backup
has slowed down and it goes on. It
slows down while it is taking the backup of the mounted
Linux server filesystems. It even doesn't give any error
messages. I went to the syslog and found the following

vmunix: NFS getattr failed for server tc4hq: RPC: Authentication error
vmunix: NFS fsstat failed for server tc4hq: RPC: Authentication error

Can anyone please help me to solve this problem.

Regards
Rajesh

[Peter Godron]

Rajesh,
assuming the backup still works, has anything been changed on the network (additional hubs etc.)? Are there network failures on the network card? I assume the Linux box is not using heavily using the disks you are trying to back up?
Regards

[Rajesh R_2]

Peter,

No network problems were there. Everything is fine. The two servers are connected to the same switch. Earlier the load on the Linux box was more. Now it has reduced very much.

[Ravi_8]

Hi,

check for NFS services on both the machines, restart NFS if needs

[Rajesh R_2]

Ravi,

I have done all this but still the problem exists

[Stephen Keane]

Are any of the users on the problem server in more than 8 groups?

[Rajesh R_2]

Stephen,

No user is in more than 2 groups

[Ralph Grothe]

Do you have to NFS mount the filesystems you need to backup at all?
I can imagine this RPC stuff being far too much overhead that unnecessarily could slow things down.
Could it be an option to you to simply dump the Linux filesystem remotely via ssh/rsh and pipe the dump into your tape device?
Or if you don't want an ext2/3 dump on your tape (or use other filesystems like ReiserFS or XFS), you could use classic tar/cpio pipes
(n.b. GNU tar on your Linux box offers the -l flag to restrict the archive to one filesystem)
If your Linux host was setup with filesystems on LVM you could even temporarily create an LVM snapshot.

[Stephen Keane]

What do the following give

rpcinfo -s server_a
rpcinfo -s server_b

rpcinfo -m server_a
rpcinfo -m server_b

where server_a and server_b are your server names (in no particular order).

[Rajesh R_2]

rpcinfo -s server_a(HP-UX)

program version(s) netid(s) service owner
100000 2,3,4 udp,tcp,ticlts,ticotsord,ticots rpcbind superuser
100024 1 udp,tcp status superuser
100021 2,4,3,1 udp,tcp nlockmgr superuser
100020 1 tcp,udp llockmgr superuser
100099 1 ticotsord,ticots,ticlts automountd superuser
100005 3,1 tcp,udp mountd superuser
100003 3,2 tcp,udp nfs superuser



rpcinfo -s server_b(LINUX Box)
There is no -s option on Linux

rpcinfo -m server_a

PORTMAP (version 2) statistics
NULL SET UNSET GETPORT DUMP CALLIT
0 0/0 0/0 6/7 0 0/0

PMAP_GETPORT call statistics
prog vers netid success failure
nfs 3 tcp 1 0
nfs 2 tcp 1 0
nfs 0 tcp 1 0
10003 0 tcp 0 1
mountd 1 tcp 1 0
status 1 udp 2 0

RPCBIND (version 3) statistics
NULL SET UNSET GETADDR DUMP CALLIT TIME U2T T2U
0 0/0 0/0 0/0 2 0/0 0 0 0

RPCBIND (version 4) statistics
NULL SET UNSET GETADDR DUMP CALLIT TIME U2T T2U
0 59/59 113/113 0/0 0 0/0 0 0 0
VERADDR INDRECT GETLIST GETSTAT
0 0 0 2


rpcinfo -m server_b(LINUX Box)
There is no -s option on Linux


Some error messages in syslog are also given below

automountd[3507]: mount_get_devid: server tc4hq not responding for /hqvar
automountd[3507]: server tc4hq:/var/spool/mail is down and can not get
the device id! errno= 238


Regards
Rajesh