NFS Server for a AIX system does not work

[Thomas Schlimm]

Hi all,
I try to use my HP-UX box as a NFS server for other unix systems.
This do not work. I get a "time out" message.

After several analysis activities I found out, that the problem is the RPC service "rpcbind".

After starting the rpcbind in debug mode (-d) on the HP-UX box I try to get a "rpcinfo -p " from the AIX box.
I can see immediatly the following on the HP-UX server in the debug log from the rpcbind:
rpcbind: check_rmtcalls: rpcb_rmtcalls == 0

After approx. 30 secondes I get the following on the AIX client:
rpcinfo: can't contact rpcbind: : RPC: 1832-008 Timed out

Any ideas?

Regards,
Thomas.

[Florian Heigl (new acc)]

Can You provide us with some more debug information?

- does rpcinfo -p localhost work on the AIX host
- does rpcinfo -p localhost work on the HP-UX host (I assume yes)
- Output of following command on the AIX host:
nfso -a
; nfso -r -a
- try setting nfso -o nfs_use_reserved_ports=1 (I DO NOT recall the exact syntax, sorry.

[Steven Protter]

Shalom,

I would very quickly check the /etc/exports and perhaps the /var/adm/inetd.sec file(probably not).

the /etc/exports file can easily contain logic on the NFS server that does not permit certain hosts or ip addresses to connect to the NFS share.

SEP

[Thomas Schlimm]

Hi Florian,

thanks for the quick answer.

The "rpcinfo -p localhost" works correct on both machines.
I attached the output from the nfso-command after changing the parameter "fs_use_reserved_ports". The change of the parameter was without a positive reaction.

I wonder if there is a problem with name resolution on the nfs server box.
I have no name server in the net.
The boxes are separated in different vlans.
I checked the nsswitch.conf and the resolv.conf.

[Thomas Schlimm]

Hi Steven,

thanks for the answer.

The issue cannot rely on the /etc/exports because I get the error even if I just use the rpc services.
The /var/adm/inetd.sec cannot be involved because in HP.UX 11i the rpcbind is started as a service and will not start with inetd.

Regards,
Thomas.

[Florian Heigl (new acc)]

Can You do some more testing?

Gather the rpcbind registration number via rpcinfo like in the following example:
program vers proto port service
100000 4 tcp 111 rpcbind
100000 2 tcp 111 rpcbind
100000 4 udp 111 rpcbind
100000 2 udp 111 rpcbind

and then try accessing it's different versions from AIX via UDP and TCP.

( I think we agree that on one of the two hosts the rpcbind is not working as expected.)

[Victor BERRIDGE]

Hi Thomas,
What version of aix are you running?
On 4.3.3 it can be tricky but on 5 it is transparent...


All the best
Victor

[Thomas Schlimm]

Hmm, the situation is very strange.
I tried to use a SuSE Linux as the NSF client.
The NFS-server on the HP-box works fine.

Therefore I think there is a problem on the implementation of NFS in AIX.

I tried the NFS client on AIX 5.2 and AIX 5.3.

[Victor BERRIDGE]

Hi Thomas,
I just tried :
HPUX10.20 to aix4.3.3
HPUX11.11 to aix5.1

All fine I will look if I can find a aix5.2...


All the best
Victor

[Victor BERRIDGE]

Hi again,
HPUX side:
"/etc/exports" 10 lines, 670 characters
octant:/etc $ exportfs -a
octant:/etc $ uname -a
HP-UX octant B.11.11 U 9000/800 168921511 unlimited-user license
octant:/etc $ exportfs|grep /sm/export
/sm/export -rw=omega:draco:deimos2,root=omega:alphard:draco,access=alphard:pallas:hooke1:hooke2:hooke3:caph:deimos:mairan:abenezra:briggs:street:tnvip2
octant:/etc $

AIX side:
tnvip2:/ $ mount octant:/sm/export /mt
tnvip2:/ $ uname -a
AIX tnvip2 2 5 00CCFB4C4C00
tnvip2:/ $ oslevel
5.2.0.0
tnvip2:/ $ df -k
Filesystem 1024-blocks Free %Used Iused %Iused Mounted on
/dev/hd4 131072 103696 21% 3060 5% /
/dev/hd2 2883584 630804 79% 60469 9% /usr
/dev/hd9var 262144 234396 11% 577 1% /var
/dev/hd3 262144 160172 39% 1005 2% /tmp
/dev/fwdump 262144 253668 4% 18 1% /var/adm/ras/platform
/dev/hd1 131072 126348 4% 176 1% /home
/proc - - - - - /proc
/dev/hd10opt 131072 43024 68% 1532 5% /opt
octant:/sm/export 5242880 306716 95% 10513 12% /mt
tnvip2:/ $

Is there something you me to check for you?
Have another check of the /etc/exports file and reissue the export (exportfs -a)
And try stop/start all the TCPIP group subsystem but start by stop/startsrc -g nfs

I would try to start by
stopsrc -s rpc.lockd and rpc.mountd
then start them again
Its a long time since I touched seriously aix so consider this as just thoughts...

Good luck
Victor