PWIP Driver (for DECnet-Plus and PATHWORKS) fails to start

[Petr X]

Hello,
I've installed on our enviroment a TCPIP patch kit v5.4 ECO6 (OVMS ver. 7.3-2). Installation went smooth till I discovered I'm no longer able to get connected via "set host" command. After short investigation I found this issue:

type $1$DGA100:[SYS0.SYSMGR]TCPIP$PWIPACP_SMFT71.LOG

Tue Mar 10 10:50:27 2009: getLocalAddr : getaddrinfo failure: 2

Tue Mar 10 10:50:27 2009: interfaceInit : Could not getLocalAddr!

Tue Mar 10 10:50:27 2009: main : PWIPdriver, PWIPacp interface init failure, aborting PWIP ACP startup...


I've been searching for a solution and found that resolver should be configured, but this is not a option in this case. I need to find a workaround solution for that case.

[Steven Schweda]

I know nothing, but "Could not getLocalAddr!"
sounds to me like a name-to-address look-up
problem, so I'd look at the local HOSTS data
base and/or DNS to see what's there for this
system. (I wouldn't expect a TCPIP ECO to
change anything important in this
neighborhood, but what do I know?)

> [...] TCPIP patch kit v5.4 ECO6 [...]

Why 6? Around here:

alp $ tcpip show version

HP TCP/IP Services for OpenVMS Alpha Version V5.4 - ECO 7
on a COMPAQ Professional Workstation XP1000 running OpenVMS V7.3-2

[Petr X]

Hi Steven,
Thanks a lot for your response.
Regarding my case, I've been looking to some configs and I found this:

1) I execute "MC NCL" and than
NCL>SHOW ALL

....
[ DNA_CMIP-MICE ] ,
[ DNA_SessionControlV2 , number = 19 ] ,
[ DNA_OSItransportV1 , 'DEC0'H ] ,
...
[ DNA_IP , 0.0.0.0 ] <-- THIS IS STRANGE
....

The line I've marked with the arrow is different on other systems like in this example:

....
[ DNA_CMIP-MICE ] ,
[ DNA_SessionControlV2 , number = 19 ] ,
[ DNA_OSItransportV1 , 'DEC0'H ] ,
[ DNA_IP , 10.162.210.66 (XXXXXX) ]
....

Which points me to question why is not an IP assigned here for the first case???

Cheers
Petr

[Steven Schweda]

> [...] <-- THIS IS STRANGE

Perhaps for you. Mine looks like that, too.
(But I don't use DECnet-over-IP, or PW, or,
...)

If I were looking for IP info, NCL is not
where I'd go. What does "TCPIP SHOW HOSTS"
say? "TCPIP SHOW INTERFACE"?


> [ DNA_IP , 10.162.210.66 (XXXXXX) ]

If you hide enough details (like "XXXXXX"),
you can make it impossible to puzzle out
what's actually happening. With an address
like 10.x.x.x, how much secrecy do you
believe that you need?

For a good time, I just ran
SYS$MANAGER:TCPIP$CONFIG.COM, and asked it to
"Configure PWIP Driver" (Start and enable).
_Now_, I get this from NCL:
[ DNA_IP , 209.98.249.184 ]
instead of this:
[ DNA_IP , 0.0.0.0 ]
which I was getting. Perhaps you should give
that a try.

[Petr X]

Yes, I did the "@SYS$MANAGER:TCPIP$CONFIG.COM" procedure already before, still the result is presented in this output:

FT71:SYSTEM> ty $1$DGA100:[SYS0.SYSMGR]TCPIP$PWIPACP_SMFT71.LOG
Wed Mar 11 17:20:15 2009: getLocalAddr : getaddrinfo failure: 2
Wed Mar 11 17:20:15 2009: interfaceInit : Could not getLocalAddr!
Wed Mar 11 17:20:15 2009: main : PWIPdriver, PWIPacp interface init failure, aborting PWIP ACP startup...
FT71:SYSTEM>

so we are back to square one :(


Regarding your point "info about IP" I've not been looking for IP in general, but more likely I tried to point out to a strange behaviour regarding first and second case.

[Bill Hall]

Petr,

I would run TCPIP$CONFIG and review or even reconfigure all of the core environment configurations. This may just fix the issue.

I think I agree with Steve, that this is probably a local host file "corruption". Running through the core environment configuration may fix the problem. Then again, you might have to manually set nohost and completely reconfigure this host.

The NCL output is just a red herring, you won't see the correct tcpip address in the DNA_IP tower until you have successful incoming or outgoing DECnet over IP traffic. You can't get any traffic if the PWIP driver doesn't start.

Bill

[Steven Schweda]

> so we are back to square one :(

We might get to square two if we could see
your TCPIP SHOW reports.

> [...] I did the
> "@SYS$MANAGER:TCPIP$CONFIG.COM" procedure
> [...]

And you think that that says enough about
what you did to be useful?

[Volker Halle]

Petr,

what does UCX SHOW HOST SMFT71 report ? It must be a DNS entry in the local host database. Is it ?

Does the IP address match UCX SHOW INT (any of the IP adresses reported) ?

Volker.

[Petr X]

Hi All, thanks a lot for your responses, I appreciate them a lot.

uxc sh ho shows:
...
10.163.201.65 SMFT71, SMFT71_O
...

FT71:SYSTEM> UCX SHOW INT
Packets
Interface IP_Addr Network mask Receive Send MTU

IE0 10.0.1.1 255.255.255.0 0 1 1500
IE2 10.15.15.1 255.255.255.224 1137 59 1500
IE6 10.15.15.33 255.255.255.224 1134 23 1500
LE0 10.163.201.65 255.255.255.192 111818 67 1500
LE1 10.163.201.129 255.255.255.128 2268628 3615614 1500
LO0 127.0.0.1 255.0.0.0 1433218 1433218 4096
WE0 10.0.0.1 255.255.255.0 0 1 1500

There is no DNS service running within the network.

[Robert Atkinson]

Could you try '$ NSLOOKUP SMFT71' ?

Rob.

[Robert Atkinson]

Also '$ UCX SHOW NAME'.

[Volker Halle]

Petr,

are there local host names defined for ALL those local IP adresses ? What local host name does UCX SHOW CONF COMM report ? Is it defined in TCPIP$HOST.DAT ?

If you check for the DNS names, always use UCX SHOW HOST/LOCAL to make sure, only the local hosts file is accessed.

Volker.

[Petr X]

Hello,
here I provide all the required info:

So, from SMFT41 to SMFT71 :

FT41:SMSC> $ NSLOOKUP SMFT71
*** Can't initialize resolver.


within the SMFT7x cluster:

FT71:SMSC> NSLOOKUP SMFT71
*** Can't initialize resolver.
=====================================
FT71:SMSC> UCX SHOW NAME

BIND Resolver Parameters

Local domain: Not defined

System

State: Started, Disabled

Transport: UDP
Domain:
Retry: Not defined
Timeout: Not defined
Servers: No values defined
Path: No values defined

Process

State: Disabled

Transport:
Domain:
Retry:
Timeout:
Servers:
Path:
FT71:SMSC> UCX SHOW CONF COMM
==============================
FT71:SMSC> UCX SHOW CONF COMM

Communication Configuration

Local host: SMFT71 Domain:

Cluster timer: 0

Interfaces: 0 Type: Default
Device_sockets: 0
Routes: 0
Services: 0
Proxies: 0

Free Maximum Minimum
Large buffers 0 0 0
Small buffers 0 0 0
IRPs 0 0
Non TCPIP buffers 0

Remote Terminal
Large buffers: 0
UCBs: 0
Virtual term: disabled
==================================
FT71:SMSC> UCX SHOW HOST/LOCAL

LOCAL database

Host address Host name

....
10.163.201.65 SMFT71, SMFT71_O
10.163.201.129 SMFT71, SMFT71_M
10.15.15.1 SMFT71, SMFT71_S1
10.15.15.33 SMFT71, SMFT71_S2
10.0.0.1 SMFT71, SMFT71_I1
10.0.1.1 SMFT71, SMFT71_I2
10.163.201.66 SMFT72, SMFT72_O
10.163.201.130 SMFT72, SMFT72_M
10.15.15.2 SMFT72, SMFT72_S1
10.15.15.34 SMFT72, SMFT72_S2
10.0.0.2 SMFT72, SMFT72_I1
10.0.1.2 SMFT72, SMFT72_I2
10.163.201.67 SMFT73, SMFT73_O
10.163.201.131 SMFT73, SMFT73_M
10.15.15.3 SMFT73, SMFT73_S1
10.15.15.35 SMFT73, SMFT73_S2
10.0.0.3 SMFT73, SMFT73_I1
10.0.1.3 SMFT73, SMFT73_I2
10.163.201.68 SMFT74, SMFT74_O
10.163.201.132 SMFT74, SMFT74_M
10.15.15.4 SMFT74, SMFT74_S1
10.15.15.36 SMFT74, SMFT74_S2
10.0.0.4 SMFT74, SMFT74_I1
10.0.1.4 SMFT74, SMFT74_I2
10.163.201.69 SMFT75, SMFT75_O
10.163.201.133 SMFT75, SMFT75_M
10.15.15.5 SMFT75, SMFT75_S1
10.15.15.37 SMFT75, SMFT75_S2
10.0.0.5 SMFT75, SMFT75_I1
10.0.1.5 SMFT75, SMFT75_I2
10.163.201.70 SMFT76, SMFT76_O
10.163.201.124 SMFT76, SMFT76_M
10.15.15.6 SMFT76, SMFT76_S1
10.15.15.38 SMFT76, SMFT76_S2
10.0.0.6 SMFT76, SMFT76_I1
10.0.1.6 SMFT76, SMFT76_I2
...

I've executed this cmd by the meantime:

FT71:SYSTEM> @sys$manager:net$configure advanced

....
Configuration Options for Node SMFT71

[0] Exit this procedure

[1] Perform an entire configuration
[2] Change naming information
[3] Configure Devices on this machine
[4] Configure Transports
[5] Configure Timezone Differential Factor
[6] Configure Event Dispatcher
[7] Configure Application database
[8] Configure MOP Client database
[9] Configure Cluster Alias
[10] Replace MOP Client configuration
[11] Configure satellite nodes
[12] Configure cluster script locations

* Which configuration option to perform? [1] : 2
* Enter the directory services to use on the system [LOCAL,DOMAIN] :
* Enter the full name for directory service LOCAL [LOCAL:.HOME.SMFT71] :
* Enter the fully qualified host name for DNS/BIND [SMFT71.FT] :
* What is the synonym name for this node? [SMFT71] :
* Naming cache timeout value? [30-00:00:00] :
* Session Control Address Update Interval ? [10] :
* Naming cache checkpoint interval? [08:00:00] :
%RENAME-I-RENAMED, SYS$SYSROOT:[SYSMGR]DECNET_REGISTER_IMPORT_FILE_SMFT71.TXT;1 renamed to SYS$SYSROOT:[SYSMGR]DECNET_REGISTER_IMPORT_FILE_SMFT71.TXT-OLD;3
%NET$CONFIGURE-I-IMPORTFILECREATED, created the DECNET_REGISTER import file

============================

also checked this file:

FT71:SMSC> ty SYS$SYSROOT:[SYSMGR]DECNET_REGISTER_IMPORT_FILE_SMFT71.TXT-OLD;1
=Type DECnet Node
=Version 1

=LocalFile Name Template LOCAL:.HOME.*

=DECdns Name Template ???
=DECdns Synonym Directory ???
=DECdns Reverse Address Directory ???

SMFT71 SMFT71 -
FT71:SMSC>

and the latest one:

FT71:SMSC> ty SYS$SYSROOT:[SYSMGR]DECNET_REGISTER_IMPORT_FILE_SMFT71.TXT;1
=Type DECnet Node
=Version 1

=LocalFile Name Template LOCAL:.HOME.*

=DECdns Name Template ???
=DECdns Synonym Directory ???
=DECdns Reverse Address Directory ???

SMFT71 SMFT71 -
FT71:SMSC>

...but still I'm missing the point what might be wrong. anyway thanks again to all who responds to my issue, mostly appreciated!

[Paul Nunez]

Hi,

There is a known problem with one of these tcpip versions where pwip simply will not start if the BIND Resolver is not enabled/configured.

Look for an update (it's long been fixed) since you indicate you can't enable the BIND Resolver...

HTH,

Paul

[Robert Atkinson]

Petr, is there a reason you can't started the Bind Resolver? You can simply point it at your DNS servers so the requests are forwarded on.

Rob.

[Petr X]

Hi Robert,
yes, the reason is, there is no DNS no NAMESERVERS running on this enviroment, which disables me to use resolver and be happy :). I'll try to seek for the update Paul suggested.

Hello Paul, I'm just askin', don't you have a direct link to this fix?

Thanks in advance for your kind response.

[Robert Atkinson]

Petr, there's an ECO 7 kit at ftp://ftp.itrc.hp.com/openvms_patches/alpha/V7.3-2/

Even though you don't have an DNS servers, you could/should still start the Bind resolver. That will probably get rid of the problem.

Rob.

[Petr X]

Hi Robert,
thanks for your resposne, I'm also aware of ECO7 kit. This option is discussed with our customer.

I'd like to add to my issue new points.
We have identified that on other systems there is a "DNVOSIECO03" patch installed, on SMFT71 not. there was only ECO2 for Decnet installed. I've decided to apply this patch to have this releases equal and eliminate this diference. The result was not successfull. I'm still seeking for a fix Paul mentioned, but so far without success.

To the point with resolver part. I understand that resolver could easily deal with this, but customer does not approve to start this service.

[Paul Nunez]

Hello again,

Finally found a few free cycles to look into the pwip / dns bind resolver issue I referred to and found it was present in v5.4 eco2.

I'm not certain the fix is any subsequent eco kit (based on Petr's post, it appears the fix was not included or is no longer present - a regression). IIRC, the customer I assisted (several years ago ;o)reverted to the previous (prior to eco2) pwip images to fix it.

I've got a server running OVMS 7.3-2 and TCP/IP v5.4 ECO4 that I'll disable bind on tomorrow and see if it breaks pwip. If not, I'll upgrade to eco6 and retest.

Or you could enable the bind resolver - just temporarily - to determine if I'm leading you down the primrose path or we're actually on to the problem...

Paul

[Petr X]

Hi Paul,
thanks a lot for your response suggestion. If I do enable resolver, how should I configure it if there is no DNS no Nameservice running within the network ?

Petr

[Ian Miller.]

Trey enable the resolver and point it at localhost as its DNS server

[Petr X]

Hello everyone,
I'd like to thank you all for your responses, support suggestions and help.
The solution was provided by all of you very soon but the missing point was my lack of knowledge. I've finally tried to configure resolver and point it to localhost name SMFT71. After that reconfiguration + reboot thinks start to behave as normal.

Thanks again a lot to all!

[Paul Nunez]

Hi Petr,

Glad it's working. Not so glad, however, that enabling the BIND Resolver may have fixed it.

I did a bit more research and found the fix was put in ECO3 (it's in the ECO4 release notes :O).

Thus far I've tried reproducing with ECO4 and could not (but I think I need to reboot the server and let TCP/IP start fresh with the Bind Resolver disabled to be 100% sure).

Haven't had a chance to try eco6 (nor eco7) yet.

Wonder if just the reboot could have fixed it.?.

Paul

[Paul Nunez]

Hi,

Installed eco6 and now TCPIP$PWIPACP won't start - exact same errors as Petr posted.

I'll try ECO7...

Paul

[Paul Nunez]

Hi,

It's working again after installing the EC0 7 kit...

Faith (temporarily) restored.

Paul