Cross Platform SAN problem

[MoreOger]

Hi

we are running a cross platform SAN in a test environment. It consists of various NW6 and W2k servers. We installed ArcServe 9 on all machines considering the document

http://support.ca.com/techbases/basb9nw/basb9nw_1015.html.

Now we have the problem that the primary W2k server couldn't initialize the tape library.

The CAPEER.LOG on the Primary NW machine shows the following error messages:

---begin capeer.log---
(09/29 12:31:16) The peer on [BACKUPW2K1] is now active.

(09/29 12:31:21) Sending initialization to the requesting server [BACKUPW2K1]

(09/29 12:31:21) Waiting to get lock for initilization/enumeration

(09/29 12:31:21) Server [BACKUPW2K1] Requesting semwait

(09/29 12:31:21) ESEMWAIT : out of esemwait

(09/29 12:31:21) Enumeration has started for client 0 on server [BACKUPW2K1]

(09/29 12:31:21) Received Data Type 0 from Client

(09/29 12:31:21) (Receive Update) Error encountered during attempt to connect:

(09/29 12:31:21) (**RPC**) [BACKUPW2K1: RPC: Name to address translation failed - n2a: hostname not found]

(09/29 12:31:21) Recover: An error occured while attempting to connect to [BACKUPW2K1]

(09/29 12:31:21) Recover: Enterprise Discovery is attempting to correct the error.

(09/29 12:31:21) Recover: Attempting to 'ping' the server...
(09/29 12:31:21) Recover: Address translation failure.
---end capeer.log---

It seems to be a name resolution problem, but we can ping every server in the SAN by name from all servers in the SAN.

Any hints or solutions?

 

[davidmichel]

First thing to do is remove the cross platform support. Work only with the NT system and get them working. Once they are working on their own then bring Cross Platform back into it.

Oh one real important item is to make sure on the Windows side the library and tape devices are disabled in the Windows Device Manager and the system restarted. This needs to be done on each of the windows systems running ARCserve. Those OS drivers can cause real problems in a SAN configuration.

 

[MoreOger]

Thx for the hint.

I installed Arcserve again on both sides (Netware and Windows). Now we encountered the following problem.

The NW primary can see the library and is able to manage it.

We configured the SAN environment as followed

BACKUPNW
+BACKUPNT

So the NT machine is a distributed server to the NW primary.
In this config the NT machine can't see the library ("Tape engine initializing ..." until timeout).

When we remove the NT machine from SAN-configuration (NW is kept as primary server) the NT can see the library but shows also the following message:

This is a shared device on a distributed server and tapeengine is unable to connect to primary server. Please make sure, that tape engine is running on the primary server.

Further hints or solutions?

 

[MoreOger]

Further information:

We found an error in the edf.ini on BACKUPNT. The sections [MemberName] and [EnterpriseSetup] were in wrong order. We moved the [EnterpriseSetup] section to the beginning of the file. Now the CAPEER.LOG shows the following errors:

(10/03 16:16:50) CAPEER: Local peer has loaded and is begining intiailziation.
(10/03 16:16:50) Client connection mode has been set to normal.
(10/03 16:16:50) Contacting CATIRPC to establish the PCS RPC services
(10/03 16:16:50) CAPEER: Loading the Shared Domain Server List into memory
(10/03 16:16:50) Constructing the list of enterprise Members from the EDF **
(10/03 16:16:50) serverName [BACKUPNW1] running on platform [NETWARE] found in config file
(10/03 16:16:50) Attempt made to get server address in unknown protocol
(10/03 16:16:50) Failed to obtain IPX address for [BACKUPNW1] from NetWare.
(10/03 16:16:50) serverName [BACKUPW2K1] running on platform [NT] found in config file
(10/03 16:16:50) Total number of server found in config file = [2]
(10/03 16:16:50) Total number of server found with same platform as host (including host) = [1]
(10/03 16:16:50) Total number of server found with different platform as host = [1]
(10/03 16:16:50) The load time behavior was not in the EDF, default to static
(10/03 16:16:50) Static behavior has been directed by the EDF for load time sempahore location.
(10/03 16:16:50) The fault behavior was not in the EDF, default to static
(10/03 16:16:50) Enterprise sempahore control is in static mode.
(10/03 16:16:50) Protocol "tcp" has been specified in the EDF
(10/03 16:16:50) CAPEER: Peer services have been established, beginning network discovery
(10/03 16:16:50) Begining initial enterprise state discovery.--**--**--**--
(10/03 16:16:50) Attempting to contact the peer on server [BACKUPNW1]
(10/03 16:16:50) UNABLE to communicate with the Primary PEER.
(10/03 16:16:50) (*Ping Failure*): RPC service is unavailable or server [BACKUPNW1] is down.
(10/03 16:16:50) RPC services are not available on [BACKUPNW1] or the server is down.
(10/03 16:16:50) Attempting to contact the peer on server [BACKUPW2K1]
(10/03 16:16:50) The peer on [BACKUPW2K1] is now active.
(10/03 16:16:50) Enterprise initial discovery is complete.--**--**--**--
(10/03 16:16:50) (Static) The ghost will be loaded on [BACKUPNW1].
(10/03 16:16:50) Waiting on notification of ghost ready
(10/03 16:18:31) Primary PEER is present and functioning.

We are able to ping BACKUPNW1 without problems.

Hints or solutions ?

 

[Gery]

Hi
Have a look if CATIRPC.NLM is loaded on NW server. Must be loaded. I suppose NT node is a Primary Server too, which is placed as Distributed under the NW Primary Server?
Kind regards
Gery