recievng NCP error upon reboot

[Lucie]

DECNET PHASE 4 - on OpenVMS I64 8.3.

I am not sure why I am getting these errors on reboot. Network is fine. Any suggestions?

NCP-W-OPEFAI, Operation failure

Message from user DECNET on N058V6
DECnet starting

%SYSTEM-F-ILLIOFUNC, illegal I/O function code
%NCP-W-OPEFAI, Operation failure

%SYSTEM-F-ILLIOFUNC, illegal I/O function code
%NCP-W-UNRCMP, Unrecognized component , Circuit
Circuit = EWA-0

%NCP-W-UNRCMP, Unrecognized component , Circuit
Circuit = EWA-1

%NCP-I-NOINFO, No information in database
%%%%%%%%%%% OPCOM 22-MAR-2010 13:55:52.01 %%%%%%%%%%%


Later in the reboot process DECNET is started

%TCPIP-I-LOADSERV, loading TCPIP server proxy information
%%%%%%%%%%% OPCOM 22-MAR-2010 12:13:43.68 %%%%%%%%%%%
Message from user DECNET on N058V6
DECnet event 4.10, circuit up
From node 2.101 (N058V6), 22-MAR-2010 12:13:40.60
Circuit LLA-0

[EdgarZamora_1]

It looks like you have a LAN Failover set defined. I haven't used it with DECnet but I would assume you will need to reconfigure your DECnet to remove the use of the EWA circuits.

[Robert Gezelter]

Lucie,

More information about your DECnet configuration would be helpful.

In the interim, I am more or less in agreement with Edgar. I suspect that you may not have accessible EWA-0 and EWA-1 devices.

As a start perhaps a LIST KNOWN CIRCUITS and a LIST KNOWN LINES from within NCP and a SHOW DEVICES from DCL would be helpful.

- Bob Gezelter, http://www.rlgsc.com

[Lucie]

Edgar, I thought I would need to clear EWA0/EWA1, These error have been going on for a while, I just happen to catch the console output when I was rebooting.

Bob, Here is the info- I didn't include the disks, I thought you wanted to see the ewa, eia and lla* devices
[SYSMGR]$MC NCP LIST KNOWN CIRCUITS


Known Circuit Permanent Summary as of 22-MAR-2010 16:19:29

Circuit State

EIA-0 off
EIA-1 off
EWA-0 on
EWA-1 on
LLA-0 on
[SYSMGR]$MC NCP LIST KNOWN LINES


Known Line Permanent Summary as of 22-MAR-2010 16:19:34

Line State

EIA-0 off
EIA-1 off
EWA-0 on
EWA-1 on
LLA-0 on
[SYSMGR]$MC NCP SHOW KNOWN CIRCUITS


Known Circuit Volatile Summary as of 22-MAR-2010 16:19:43

Circuit State Loopback Adjacent
Name Routing Node

EIA-0 off
EIA-1 off
LLA-0 on 2.623
[SYSMGR]$MC NCP SHOW KNOWN LINES


Known Line Volatile Summary as of 22-MAR-2010 16:19:50

Line State

EIA-0 off
EIA-1 off
LLA-0 on
[SYSMGR]$SHOW DEV

Device Device Error
Name Status Count
OPA0: Online 0
ASN0: Online 0
FTA0: Offline 0
RTA0: Offline 0
RTB0: Offline 0
TNA0: Online 0
TNA3: Online 0
TTA0: Online 0
TTB0: Online 0

Device Device Error
Name Status Count
$1$GGA900: Online 0
$1$GGA901: Online 0
$1$GGA902: Online 0
$1$GGA9003: Online 0
CMA0: Online 0
EIA0: Online 0
EIA2: Online 0
EIA3: Online 0
EIB0: Online 0
EIB2: Online 0
EIB3: Online 0
EWA0: Online 0
EWA5: Online 0
EWB0: Online 0
EWB3: Online 0
FGA0: Online 0
FGB0: Online 0
GHA0: Offline 0
LLA0: Online 0
LLA3: Online 0
LLA4: Online 0
LLA5: Online 0
LLA6: Online 0
LLA7: Online 0
LLA8: Online 0
MPA0: Online 0
OHA0: Online 0
OHB0: Online 0
PKA0: Online 0
PKB0: Online 0
PPP0: Online 0
SEA0: Online 0
SRA0: Online 0
UCM0: Online 0
USB0: Online 0
WSA0: Offline 0
[SYSMGR]$

[The Brit]

Lucie,
Question! Is your failover set LLA0 made up of EIAn devices, or EWAn devices??

If LLA0 = EWA0 + EWA1, then as implied by Edgar, you need to go into ncp and execute

NCP> define circuit state off
NCP> define line state off

This will remove them from the permanent database.

If they are set to "on" in the permanent data base, the system will try to bring up the lines and circuits at boot time, however since these devices are now subordinate to the LLA device, you will get the errors.

If EWA0 and EWA1 are NOT in your Failover Set, then there is some other issue.

HTH

Dave.

[Robert Gezelter]

Dave,

With all due respect I must demur from your characterization:

>NCP> define circuit state off
>NCP> define line state off
>
>This will remove them from the permanent database.

DEFINE [CIRCUIT|LINE] STATE OFF will not "remove them from the permanent database". It will prevent them from starting. PURGE [CIRCUIT|LINE] would remove the entries from the database.

- Bob Gezelter, http://www.rlgsc.com

[Lucie]

I try to determine what cards I using. I am more confused now
NCP>show know lines
Known Line Volatile Summary as of 23-MAR-2010 08:49:02
Line State
EIA-0 off
EIA-1 off
LLA-0 on
but in lancp lla0 represents ewa0 and ewb0. ewa0 link is down.

LANCP> show dev /para lla0

N058V6 Device Characteristics LLA0 (23-MAR-2010 08:55:35.10):
Value Characteristic
----- --------------
1500 Device buffer size
Normal Controller mode
External Internal loopback mode
AA-00-00-3E-91-A0 Default MAC address (Hardware LAN address)
Multicast address list
Ethernet Communication medium
FF-FF-FF-FF-FF-FF MAC address (Current LAN address)
8 Minimum receive buffers
256 Maximum receive buffers
Yes Full duplex enable
Yes Full duplex operational
AA-00-04-00-65-08 MAC address (Current LAN address)
TwistedPair Line media type
1000 Line speed (mbps)
Disabled Auto-negotiation
Disabled Flow control
Disabled Jumbo frames
"EWA0" Failover device
"EWB0" Failover device (active)
Enabled/Active Logical LAN state
0 Failover priority
Link Up Link state


the error message indicates that it does not "see" ewa0 and ewa1( there is no device ewa1 found in lancp)

[EdgarZamora_1]

You currently don't have an EWA-1 (nor an EIA-1). Maybe in the past you did? We don't know the history of your system. Currently your LAN failover set LLA-0 is composed of EWA-0 and EWB-0. The error messages you are seeing are really harmless and you can just leave things the way they are. If you want to remove EWA-0 and EWA-1 from the DECnet permanent database use the NCP PURGE LINE and PURGE CIRC commands.

[Lucie]

What I do not understand is why is ewa0 status link is down. I thought both card status should be link up but only one card is "activeâ perhaps ewa0 has gone bad?

I will purge ewa-1 and eia-1 from database

FYI, As far as I can tell - these errors have been going on since 4/2009.

[Volker Halle]

Lucie,

when building a LAN Failover device (LLA0 in your case), the underlying physical LAN interfaces (in your case EWA0 and EWB0) are not available to other components of OpenVMS, only to the LAN failover driver (SYS$LLDRIVER). This explains, that DECnet cannot start/find lines and circuits EWA-0 and EWA-1, which correspond to the VMS devices EWA0 and EWB0.

The message is harmless, but it should still be eliminated by correctly PURGING EWA-0 and EWA-1 from the DECnet lines & circuits database.

Volker.

[Lucie]

Volker,
If I purge EWA-0 from the database, won't it affect EWA0 I thought the "names" correspond? EWA-0 = EWA0, EWA-1 = EWA1 ETC.

Also, I will need to configure LLA0 with the correct cards.

I am sorry I am confused on this, but it doesnâ t make sense. If I purge ewa-0 and ewa-1, what will the failover devices for LLA0 be? EWB0 AND EWA?

[Robert Gezelter]

Lucie,

PURGE only affects the configuration as DECnet sees it. It does not affect the underlying OpenVMS devices, which are what LLDRIVER is using.

Drawing a diagram in text is difficult, but

DECnet EWA-0 LLA-0
OpenVMS EWA0: LLA0: (composed of EWA0: and EWB0:)

- Bob Gezelter, http://www.rlgsc.com

[Volker Halle · Accepted solution]

Lucie,

EWA0 is the representation of the physical LAN interface in OpenVMS.

EWA-0 is the DECnet line and circuit, which is controlling and running the DECnet protocols on top of EWA0.

Same explanation for EWB0 and EWA-1.

You have configured EWA0 and EWB0 into a LAN failover set (LLA0) and you've configured DECnet to run over this LAN failover set (DECnet Circuit/Line LLA-0).

The DECnet configuration procedure (NETCONFIG.COM) is not 'clever' enough to recognize, that EWA0 and EWB0 are part of a LAN failover set and therefore NOT AVAILABLE for other uses.

Please PURGE and CLEAR DECnet LINEs and CIRCUITs EWA-0 and EWA-1 and all these error messages will disappear. And you DECnet configuration will be correct and clean.

If you don't use EIA0 and EIB0 (EIA-0 and EIA-1 respectively) for running DECnet, you may also want to consider to PURGE/CLEAR those circuits/lines.

Volker.

[The Brit]

My apologies to Bob, he is of course correct. "Define" will not remove the entries from the permanent db.

However, I would still suggest that setting the Lines and Circuits to "State = OFF" in the permanent DB, would achieve the same result of getting rid of the error messages.

While I agree that running Purge/Clear to actually remove EWA-0 and EWA-1 is an OK thing to do, as I recall it does make things a little more complicate if the Failover Set is subsequently disolved, since the circuits and lines would have to be recreated in NCP.

If I am wrong, please say so. I'm always looking to tighten my understanding of these things.

Dave.

[Lucie]

I purged and cleared the lines and circuits to EWA-0 AND EWA-1 before The Britâ s comments (I should have turned them off) I rebooted the system and did not receive any errors.
I am still confused about DECNET vs. LAN failover. Is there a good manual or whitepaper that I may read to get a better handle on DECnet and LAN failover?

I appreciate everyone indulgence with in my lack of knowledge with DECnet and LAN failover configurations.

[Robert Gezelter]

Dave,

No apology is necessary.

My main suggestion here is to avoid creating confusion if someone else has to look at things in the future.

If I were in Lucie's situation, I would be strongly tempted to either use DEFINE to set state to OFF (as you suggested), OR to PURGE the definitions.

Generally, my inclination is to PURGE. This is not a temporary situation. I have seen sites with many lines defined, for long defunct configurations, that no individual on the site has any recollection of. It makes sorting out problems difficult.

What I would recommend is perhaps creating a command file with the NCP commands to define the LINE/CIRCUIT configuration, with full comments (in NCP, lines beginning with "!"; admittedly this is lightly documented).

Example:

! LLA-0 is presently composed of EWA-0 and EWA-1
! used as a LAN Failoverf set
! Changes to hardware configuration require
! corresponding changes to the failover set
! using LANCP.
DEFINE CIRCUIT ...
DEFINE LINE ...

The automatic creation of NCP commands by NETCONFIG.COM is useful, but complex configurations can out-distance its utility.

- Bob Gezelter, http://www.rlgsc.com

[Volker Halle]

Lucie,

LAN Failover is described in the 'HP OpenVMS System Manager's Manual, Volume 2' in chapter 10 in the sub-chapter 'Understanding LAN Failover'

http://h71000.www7.hp.com/doc/82final/aa-pv5nj-tk/aa-pv5nj-tk.html

In short: you create a LAN Failover Set (LL device) from physical LAN interfaces. Only one of these LAN interfaces is active at any time. If the link of the active LAN device to the switch is lost, LLDRIVER will switch to another LAN interface in this set. This is transparent to ALL procotols running on top of the LL device (IP, DECnet, SCS, LAT etc.).

ANAL/SYS and SDA> SHOW LAN shows all LAN interfaces and the protocols running on them. You'll see that there will be no procotols on EWA and EWB, all the protocols are running on LLA.

Volker.

[Ian Miller.]

Another handy command

$ MCR LANCP SHOW CONF EWA/USE

to show what protocols are using EWA