Problem with swinstall

[Phillip_11]

Hi,

I'm running 10.20 of hpux on a K200 series
system. I get the problem of with swinstall:

command: swintall -i -s /tmp/XSW800HWCR1020_10.20_800.depot -x autoreboot=true

The message I get is that RPC call to Daemon has failed could not start a managment session for seeya:/tmp/XSW800HWCR1020_10.20_800.depot". Make sure host is accessible from network and that it's daemon, swagentd is running.

I checked that swagentd is running and did notice that the dce/rpcd was not running. So I started that up by /sbin/init.d/dce start
checked the process to see that it is in fact running.

This is done on a freshly installed system. I would appreciate any other pointers. Logs is only indicating the same errors.

Another thing this is being done on the system locally so it is NOT a network issue. All the RPC daemons are up and running.

I can't install any patches until I get this working.

[Rusty Sapper]

This is just a shot in the dark, but there is a know issue on 10.20 that the measureware daemons can screw up swinstall.

do you have any measureware running?

you can stop it by running "mwa stop"

HTH

-Rusty

[John Palmer]

Have you changed the hostname and/or IP address? swinstall needs to be able to resolve the hostname.

What's the output from: uname -n

On the presumption that hostname is 'seeya', can you successfully telnet seeya ?

Regards,
John

[Sanjay_6]

Hi,

Can you do a swlist and see if it works.

you may try to restart swagentd,

/sbin/init.d/swagentd stop
/sbin/init.d/swagentd start

or

/usr/sbin/swagentd -r

Hope this helps.

Regds

[Phillip_11]

Thanks Rusty, John and Sanjay for replying to my problem.

1. I did this remotely and locally and still get the problem. So it is NOT a network issue.
As I can telnet into seeya with no problem.

2. No Measurewear is running or installed.

3. Haven't change hostname or IP address since installing the OS.

4. swlist does work. and yes I tried starting and stopping swagentd via /sbin/init.d/

After swinstalled is started by itself. It brings up the shared root paths: The reslult of that is that the RPC still fail as it could not start a management seesion for seeya:

Again I checked the swagentd which is still running. Also all the rpc daemons are running also. the swinstall.log only indicates RPC expection: Communication faiure ( dce / prc )

So I have no other clue at this point. The only thing left to do is put swagentd in debug state and see what happens. Any other suggestion?

[John Palmer]

Another possibility is having duplicate entries in /etc/hosts for your hostname with different IP addresses.

Still sounds like an IP resolution issue.

Regards,
John

[James R. Ferguson]

Hi:

Another thing to verify, is that you have a localhost entry in /etc/hosts:

127.0.0.1 localhost loopback

...and also your hostname

10.101.102.103 myhost myhost.xxx.com

Regards!

...JRF...

[Rusty Sapper]

Actually,
This is starting to sound like a problem I experienced with 10.20 a few years ago. I can't remember exaclty, but I remember having to add a line in /var/adm/sw/default.

maybe something to do with the parameter rpc_binding_info

Does this ring any bells with anyone? I remember numerous calls to the response center before I got it resolved.

-Rusty

[Anil C. Sedha]

Hi

Since you are installing from the local system you should try this

swreg -l depot depotpath

swinstall -i -s systemname:/tmpXSW800HWCR1020_10.20_800.depot -x autoreboot=true

For your RPC and network services stop and restart them

You may find them under /sbin/init.d

nfs.client
nis.client
nfs.server
nis.server(only if slave server)
nfs.core

start them in reverse order.

Regards,
Anil

[Sanjay_6]

Hi,

Can you attach the log file /var/adm/sw/swagent.log

Regds

[James R. Ferguson]

Hi Rusty:

I think what you are remembering is the "ability" for an active 'glance' or 'measureware' session to freeze 'swinstall'. You can circumvent that by adding this option to the 'swinstall':

-x rpc_binding_info=ncadg_ip_udp:[2121]

Regards!

...JRF...