Operating System - OpenVMS
1748011 Members
4541 Online
108757 Solutions
New Discussion

Re: TCPIP interfaces le0 + we1 missing after upgrade to 8.3 / ioctl failed: status = 908

 
SOLVED
Go to solution
The Brit
Honored Contributor

Re: TCPIP interfaces le0 + we1 missing after upgrade to 8.3 / ioctl failed: status = 908

Might be unrelated, however I had a similar problem when the devices were enabled for down-line loading.

 

try

LANCP>  show dev /dll

 

and if they show up as "enabled" then disable them (particularly in the Permanent DB.

 

Dave. 

mkrauss
Occasional Advisor

Re: TCPIP interfaces le0 + we1 missing after upgrade to 8.3 / ioctl failed: status = 908

Hi !

 

First I want to thank you for all the tips. We were able to reproduce the problem on our test system (ES47).  Last week we managed to get the interfaces up and running again on the test system.

We are going to implement the change in the production system in first week of August.

 

Summary:

Initially the problem seemed to be related to a missing LE0 TCP/IP interface. Digging a little bit deeper, we found, that TCPIP could not use the failover device lla0. Looking at it using LANCP, the interface was defined and created during boot, but the participating physical interfaces EWA0 and EWB0 could not be added during boot.

 

Hoff suggested, to check for a manual procecdure required during upgrade  to OpenVMS 8.3.  According to http://h71000.www7.hp.com/doc/84final/6677/6677pro_install.htmlone

we might need to convert the LANCP device database manually doing

$ LANCP

LANCP> CONVERT DEVICE_DATABASE

LANCP> SET ACP/STOP

LANCP> EXIT

$ @SYS$STARTUP:LAN$STARTUP

 

but this didn’t fix the problem.

 

Checking patchlevels of OpenVMS and TCPIP didn’t get the thing done either.

 

Dave suggested to check, whether downlineload (MOP) is enabled on the interfaces, but doing a ‘mc lancp show dev /dll’ showed, that DLL on all interfaces was disabled.

 

The best hint came from Volker Halle and an old ITRC thread I was only able to find using google cache ‘Can't start LAN failover’. Volker suggested to remove LLA0 completely and rebuilt the setup from scratch. Unfortunately, this wasn’t possible while booting in normal mode, since the physical interfaces are in use by some protocols. This can be checked by

$ ananlyze/sys

SDA> show lan

(Volker, in your post the analyze was replaced by ‘**bleep**’, I suspect you abbreviated it…)

 

We then tried to boot minimum (STARTUP_P1 = “min”) which prevented other protocols than SCA to start (SCA doesn’t harm here).

Purging and clearing lla0 from permanent and volatile lancp databases removed all reminiscences to this interface.

Now recreating the interface using

LANCP> define dev lla0/failover=(ewa0,ewb0)/ena

LANCP> set dev lla0/failover=(ewa0,ewb0)/ena

 

still produced an error

%LANCP-E-LLANACT, LAN device active, cannot be added to or removed from Logical LAN LLA0

 

but continuing with

LANCP> set device lla0 /update/vola

 

and then rebooting back to normal startup did the trick. With this reboot, LLA0 was not only created but was able to attach to a physical interface.

 

%LLA0, Logical LAN event at 12-JUL-2011 12:46:41.41

%LLA0, Logical LAN failset device created

%LLA0, Logical LAN event at 12-JUL-2011 12:46:41.42

%LLA0, Logical LAN failover device added to failset, EWA0

%LLA0, Logical LAN event at 12-JUL-2011 12:46:41.42

%LLA0, Logical LAN failover device added to failset, EWB0

%LLA0, Logical LAN event at 12-JUL-2011 12:46:41.42

%LLA0, Logical LAN failset device connected to physical device EWB0

 

The last line was the one that was missing and caused the problem.

At this point, TCPIP no longer complained about LLA0 being offline and started with its original config.

 

 Thanks again for you help.

 

Michael

 

mkrauss
Occasional Advisor

Re: TCPIP interfaces le0 + we1 missing after upgrade to 8.3 / ioctl failed: status = 908

Spoiler
 

Hi all,

 

today we wer successful in implementing the change in the production environment. Just deleting and recreating lla0 from scratch fixed the problem. It is important to do the LANCP set/def commands while booted minimum (Startup_P1 = "min"). Otherwise the physical failover devices (e.g. ewa0, ewb0) will be blocked by running protocols.

I really can't tell what has happened, but it is fixed now.

 

Best regards,

Michael