Dec-Net Plus setup

[robert70]

Hi,

I have an RX2800i running VMS8.4 with DECnet Plus installed -

I am trying to copy and backup data from an Alpha DS20E running VMS8.3 with DecNet Plus installed.

I have configure both boxes with the net$configure command as below...............

$ @net$configure
Copyright 2007 Hewlett-Packard Development Company, L.P.

        DECnet-Plus for OpenVMS network configuration procedure

    This procedure will help you create or modify the management scripts
    needed to operate DECnet on this machine. You may receive help about
    most questions by answering with a question mark '?'.

     Configuration last run by SYSTEM on 15-FEB-2012 12:47:48.35

        Configuration Options for Node EDDY

        [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

* Which configuration option to perform?                 [1] :
* Enter the directory services to use on the system  [LOCAL] :
* Enter the full name for directory service LOCAL [LOCAL:.EDDY] :
* What is the synonym name for this node?             [EDDY] :
* What type of node (Endnode or Router)?           [ENDNODE] :
* Enter PhaseIV Address                                [1.1] :
%NET$CONFIGURE-I-SCANCONFIG, scanning device configuration - please wait

        Since you are using the BASIC option, if you answer YES to
        the next question, the NSP transport will be reconfigured
        using the default values which are:

        Maximum number of logical links:            200
        Maximum Transmit and Receive Window:         20
        Maximum Receive Buffers:                   4000
        Flow Control Policy:       Segment Flow Control

        If you intend to customize the transports, then use
        the ADVANCED option instead.

* Do you want to replace the existing NSP transport script? [NO] :

        Since you are using the BASIC option, if you answer YES to
        the next question, the OSI transport will be reconfigured
        using the default values which are:

        Maximum number of logical links:            200
        Maximum Transmit and Receive Window:         20
        Maximum Receive Buffers:                   4000
        OSI applications over TCP/IP:           Enabled
        DECnet applications over TCP/IP:        Enabled
        DECnet/OSI over TCP/IP interface(s):        ALL

        If you intend to customize the transports, then use
        the ADVANCED option instead.

* Do you want to replace the existing OSI transport script? [NO] :
%NET$CONFIGURE-W-NOPWIP, DECnet over IP requires the PWIP driver to be enabled
%NET$CONFIGURE-W-NODOMAIN, DECnet over IP requires DOMAIN in the directory services list
%NET$CONFIGURE-I-EVDFND, Event Dispatcher NCL script already exists
* Replace Event Dispatcher NCL script file?             [NO] :
%NET$CONFIGURE-I-APPLICATIONFND, application NCL script already exists
* Replace application script file?                      [NO] :
* Load MOP on this system?                              [NO] :
%NET$CONFIGURE-I-MOPCLIENTFND, MOP client NCL script already exists
* Replace MOP Client script file?                       [NO] :

        Summary of Configuration

 Node Information:

        Directory Services Chosen:      LOCAL
        Primary Directory Service:      LOCAL
        Local Full name:                LOCAL:.EDDY
        Node Synonym:                   EDDY
        Phase IV Address:               1.1
        Phase IV Prefix:                49::
        DECdns Node Type:       Clerk
        Session Control Address Update Interval:        10
        Routing Node Type:              ENDNODE
        Autoconfiguration of Network Addresses: Enabled
        Routing ESHello Timer:  600
        Routing ES Cache Size:  512

 Device Information:

        Device: EIA0 (EI/82558):
                Data Link name: CSMACD-0
                Routing Circuit Name:   CSMACD-0

        Device: EIB0 (EI/82558):
                Data Link name: CSMACD-1
                Routing Circuit Name:   CSMACD-1

        Device: EIC0 (EI/82558):
                Data Link name: CSMACD-2
                Routing Circuit Name:   CSMACD-2

        Device: EID0 (EI/82558):
                Data Link name: CSMACD-3
                Routing Circuit Name:   CSMACD-3

 Transport Information:

        NSP Transport:                    Configured
           Maximum number of logical links:     200
           Maximum Transmit and Receive Window: 20
           Maximum Receive Buffers:             4000
           Flow Control Policy:                 Segment Flow Control

        OSI Transport:                    Configured
           Maximum number of logical links:     200
           Maximum Transmit and Receive Window: 20
           Maximum Receive Buffers:             4000
           OSI applications over TCP/IP:        Enabled
           DECnet applications over TCP/IP:     Enabled
           DECnet/OSI over TCP/IP interface(s): ALL

        Congestion Avoidance Disabled

 Event Dispatcher Configuration:

        Sinks:            local_sink
        Outbound Streams: local_stream
        Phase IV Relay:   Enabled

* Do you want to generate NCL configuration scripts?   [YES] :
%NET$CONFIGURE-I-MODCHECKSUM, checksumming NCL management scripts modified by NET$CONFIG
URE
%NET$CONFIGURE-I-NOTREPLACED, SYS$SPECIFIC:[SYSMGR]NET$APPLICATION_STARTUP.NCL; was not
replaced
%NET$CONFIGURE-I-NOTREPLACED, SYS$SPECIFIC:[SYSMGR]NET$EVENT_STARTUP.NCL; was not replac
ed
%NET$CONFIGURE-I-NOTREPLACED, SYS$SPECIFIC:[SYSMGR]NET$MOP_CLIENT_STARTUP.NCL; was not r
eplaced

sys$manager:net$dns_clerk_startup.ncl changed to use the new default namespace.

Your default namespace nickname is LOCAL.

Your default namespace NSCTS is 08-00-2B-0D-C0-9D-5F-FA-A9-88-43-46-95-00.

Directory Service: Local name file

Modifying the node: LOCAL:.EDDY
Modifying the node towers

%NET$CONFIGURE-I-NODERENAMED, node successfully renamed to LOCAL:.EDDY
%NET$CONFIGURE-I-FLUSHCACHE, flushing selected cache entries
%NET$CONFIGURE-I-CONFIGCOMPLETED, DECnet-Plus for OpenVMS configuration completed

        Configuration Options for Node EDDY

        [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

* Which configuration option to perform?                 [0] :
$

Eddy is the node name of the RX2800 and Dave, which was set up the same way as above,is the node name of the Alpha Server

Do I need to add in other config settings to get these 2 nodes to communicate to each other?

Copy and Backup commands not working eg from Eddy to Dave

$ copy dave::dkb1:[000000]*.* *.*
%COPY-E-OPENIN, error opening DAVE::DKB1:[000000]*.*; as input
-RMS-E-FND, ACP file or directory lookup failed
-SYSTEM-F-NOSUCHNODE, remote node is unknown

[Hoff]

Use the decnet_register tool to register the name and address of the DECnet-Plus hosts on each of the hosts.  

You'll want to read the DECnet-Plus introduction to get oriented with that product and (for a chapter with details of the decnet_register tool) the DECnet-Plus Network Management Manual from the DECnet-Plus documentation shelf. 

This decnet_register approach given you're likely not running DECdns and you're probably not connected to DECnet over IP nor to IP DNS.  But in any event, you'll need to become somewhat familiar with decnet_register.  (Put another way, there are other options for configuring the namespaces and acquiring the host names.  But for a small and unchanging network, a static configuration works fine.)

(The DECnet-Plus user interface is unfortunately far too hairly to wing-it; you'll end up reading the docs and running some web searches to get it to work.)

[robert70]

thanks steven yes the configure was done on both individual machines with seperate ip's

[robert70]

Thanks Hoff I will look into all this useful info now - I am trying to basically use Decnet Plus over TCPIP to allow backup and copy between boxes. Thanks - Will update you how I get on

[Duncan Morris]

Then don't forget to address these issues, reported by your configuration.....

%NET$CONFIGURE-W-NOPWIP, DECnet over IP requires the PWIP driver to be enabled
%NET$CONFIGURE-W-NODOMAIN, DECnet over IP requires DOMAIN in the directory services list

[robert70]

Thanks Duncan - sorted the issue with PWIP and Domain Cheers for pointing out

[robert70]

Thanks Steven - yes they are on the same Physical Network

[Hoff]

If I was looking to operate DECnet on the same network segment and if the configuration didn't have to traverse any intervening IP routers or other IP-related networking devices, then I'd likely elect to shut down DECnet-Plus, uninstall it, and install and configure DECnet Phase IV.

While there are capabilities that DECnet-Plus can provide and DECnet Phase IV cannot, DECnet Phase IV is far simpler than DECnet-Plus (that product formerly known as Phase V), and Phase IV is generally far easier to manage, and Phase IV requires far fewer resources, and Phase IV is typically also rather more stable.

Here's an intro to DECnet Phase IV management

If I had disk space for copies and such (and a more complex network), then I might well use straight ftp or sftp for the file transfers, and avoid DECnet entirely.

If this is at all related to that MSA migration you were discussing a couple of months back, then I might well discuss this migration with somebody; getting the data across is usually accomplished rather quickly.  The OpenVMS software migrations involved in porting to Itanium can take longer, though.

[robert70]

Thanks Hoff - yes this is to do with a system migration I discussed a few months back -

Have an Alpha running VMS 8.3 connected to an MSA1000 - looking to copy the data from this syste, to an RX2800i connected to a D2700 Disk Enclosure. Both Boxes on the same network.

Can transferring via FTP be done on large data sets? eg looking to transfer a total of 350Gb with the largest individual files about 4Gb. The network is 100BaseTX Full Duplex.

Cheers

[Volker Halle]

Robert,

you can use DECnet node adresses (i.e. 1.1 and 1.2 - assuming you configured the 2nd node with an address of 1.2) instead of nodenames to copy files across ($ copy filename 1.2::DISK:[dir]). I do those kinds of operations all the time when migrating old VAX or Alpha systems to the CHARON emulators.

You need to have a disk on the target system (your RX2800i), that's big enough to hold the backup/image savesets of all the disks on your DS20E or at least the biggest possible saveset.

Make sure to include $ SET RMS/NET=127/EXTEND=65535 on both nodes (e.g. in LOGIN.COM) to improve the performance of the copy over DECnet. Then write some DCL batch jobs on your DS20E to do a BACKUP/IMAGE local_disk: 1.2""disk:[dev]local_disk.BCK/SAVE for all your local disks to be copied. Depending on the speed of your LAN interconnect and the CPU power of the DS20E, don't run more than 2-3 of those jobs in parallel.

Also take special note about BACKUP-I-NOBACKUP messages. If you've set files to /NOBACKUP to save space in your daily backups, now it's time to set those files back to /BACKUP to make sure the contents gets copied over to the new machine !

Volker.

[robert70]

Thanks Volker - very useful thank you

[Hoff]

Yes, DECnet can work here.

And yes, all recent versions of ftp can be and variously are used to transfer multi-gigabyte files.   (Some of the oldest of the ftp software that's been around might toss a snit, though.)  HP uses ftp to download multi-gigabyte OpenVMS kits to customers quite regularly, after all.

100 MbE is slow, but this data-transfer operation clearly isn't time-critical.

The biggest issue with transferring larger files tends to be flaky network configurations, or flaky hardware.

One of the fastest ways to the transfer data involves adding a little extra storage hardware, and hauling stuff across on disk bricks.  Depending on the hardware that's added, you also have a backup copy, and some extra storage, and a connection for (for instance) an external SCSI tape device.   (But then this data transfer and this migration doesn't look to be time-critical.)

If using a network and ftp, then using current versions of zip and unzip (if you have extra storage to hold the intermediate files) can be used to compress the data for a faster network transfer.  zip 3.0 and unzip 6.0 (current, or later) can deal with multi-gigabyte large files.  (Older versions of zip and unzip cannot contend with larger files.)

it's also possible to zip a BACKUP saveset and ftp that, as HP does with variois of their software kits and software distributions.

If you do choose to use zip and ftp as part of this approach, then remember to use "-V" - that's a quoted, uppercase V, and the quotes and the uppercase are mandatory in the most common OpenVMS command-parsing environment - to cause zip to save the RMS file attributes.

The "gold standard" approach here is usually clustering.  Clustering allows the old and new systems to share all storage, and access all storage.  This then allows the current and new environments to operate entirely in parallel, meaning that the transitions and the migrations can be incremental.

And getting the files transferred is the smaller part of the project.

I'd expect that a brute-force transfer of a storage configuration as small as 300 GB should have been completed in a few days time to maybe a week on the outside, with that time involving setting up the DCL procedures, and waiting for the (lack of) speed of that 100 MbE connection, and then contending with whatever errors that Ethernet connection might encounter.   If all the computing hardware, storage and network hardware is correctly configured and working properly, then this transfer is probably a couple of days to implement, to transfer the data, and verify the results.  This data transfer is (usually) the easy part of a port.  Getting the applications and source code ported can sometimes take rather longer than a week.

[robert70]

Thanks Steven and Hoff -

I only have versions 2.2 of ZIP and 5.52 of UNZIP here.

Where can you pickup zip31 and unzip61 from?

Thanks

[Hoff]

Among (many) other discussions, zip and unzip intro; DCL foreign commands  has a download link, documentation on set-up and usage, and related information.  

(The versions of zip and unzip that available on the HP Freeware distros are stale, as are most of (all of?) the copies at the HP OpenVMS web site.  Use the Process Software link from the above page, or via the Info-Zip site.)

[robert70]

Hoff / Steve - thanks very much - very useful information

Will get these zip3 and unzip6 downloaded and start some testing with transferring between DS20E & RX2800i.

Plan to look at using DecNet or FTP or possibly DNFS.

Thanks

[Hoff · Accepted solution]

We're months into this host-to-host data transfer, and the configuration is to the point of testing zip and IP transfers, and this configuration doesn't have a fully-operational DECnet network.

Dealing with NFS devices for transfers and the somewhat obscure and arcane management of the VMS version of NFS is not something I'd recommend undertaking here.  (And for the IP versions you have, there are other NFS-level capacity limits that will likely be in play.)

Keep it simple.

Get the data transfered.

There is a larger project waiting with the source code port here, as well as reorganizing details such as the physical device references within existing tools and procedures.

[robert70]

all transferred - 250gb of database in about 5 hours using FTP - version with vms8.4 very rebost compared to early versions - thanks all for help