[HECnet] Multinet connection in UK

On 2018-11-04 17:21, Steve Davidson wrote: If a node is defined in both places (like a cluster member) the copy will overwrite key information that the boot node needs. This can also happen with decservers that are defined in both places. Netupdate does NOT modify nodes in the local area to prevent such problems. In fact it notifies the user that it is skipping the local area. It presumes that the local user has made their own definitions for their area. This issue has burned me bad in the past thus NETUPDATE was created. Also this can be submitted to batch. In my case it is run whenever my systems get your notification email. Pretty much automatic! -Steve Davidson Sent from my iPhone > On Nov 4, 2018, at 10:32, Johnny Billquist <bqt at softjar.se> wrote: > > Do you know in which way, and why they would be corrupted? As far as I understood things, this should only copy over node definitions from MIM, but not change anything that is not touched by the node information from MIM. So other node names should not get any troubles. > Not entirely sure about what happens to existing nodes, but I would have hoped they would still have the additional information that might have been there untouched as well. > > I do create a FIX.COM command file as well, and in the past that script was doing too much. It might still get you into trouble if you have other type of information in your local database, so I would be careful with that script. But I'm also open to suggestions for improvements for it. > > Johnny > >> On 2018-11-04 16:05, Steve Davidson wrote: >> Be careful of "ncp copy known nodes". Cluster nodes and DECserver nodes can become corrupted with this command. That is why I wrote netupdate.com >> -Steve >>>> On Nov 4, 2018, at 09:54, Johnny Billquist <bqt at softjar.se> wrote: >>>> >>>> On 2018-11-04 14:59, supratim sanyal wrote: >>>> Hi Douglas, >>>> Great, I will mark port 60004 as allocated to you. >>>> One reason for TCP is it does not require Johnny to go change the IP address on his end every time a residential ISP IP address changes. >>> >>> Which is not maybe that relevant in a link between you two... The VMS Multinet do not appear to be as flexible in configuring peers, so you might need to restart if addresses change anyway. But yes, for the RSX Multinet-compatible links, the active end can change IP address without me having to do anything on my side. TCP also have the advantage to better handle if packets might be dropped, which can be an issue over long haul links, as well as better match what these Multinet links pretend to be to DECnet, which are DDCMP links. But the VMS implementation seems to be cheating in several ways, making it look funny at times, so it might not matter much from that point. >>> >>> Also, TCP is easier if you have firewalls or NAT. UDP can also be made to work, but is usually a little more complicated here. >>> >>>> Does ?MCR NCP COPY KNOWN NODES FROM MIM TO BOTH? not work to copy over node db from MIM? >>> >>> It should work for him, once he have MIM defined. >>> >>>> I also run a neat command file by Steve Davidson that I found somewhere (maybe on MIM or STRGTE) which I am attaching ? it runs on SYS$BATCH and updates local node db automatically. The .XCOMX extension should be .COM, psilobyte rejects .COM file attachments. >>> >>> I could also mention that I have an automated system that sends mails when I update the nodename database, in case anyone would want such a thing. If you are creative, you set things up so that script is run automatically when such a mail is received. >>> >>> Johnny >>> >>>> Best, >>>> Supratim >>>> Sent from Mail <https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10 >>>> *From: *Douglas Hall <mailto:dhall.hecnet at dhcl.co.uk> >>>> *Sent: *Sunday, November 4, 2018 07:57 >>>> *To: *hecnet at Update.UU.SE <mailto:hecnet at Update.UU.SE> >>>> *Subject: *Re: [HECnet] Multinet connection in UK >>>> On Sun, 4 Nov 2018, at 11:19, supratim sanyal wrote: >>>> If you wish you can run an experiment with IMPVAX (VA, USA); I have >>>> heard trans-Atlantic loop latencies are not as bad as one might >>>> think. TCP only, Something like this in DECNET-CIRCUITS.COM should work: >>>> $ multinet set /decnet /remote=52.23.221.223 /port=60000 >>>> /device=tcpa0: /connect /tcp=connect /buffers=24 >>>> That looks to be working OK. It's very bursty to some locations, e.g. BOPOHA, but connectivity is connectivity. Thanks! Out of interest, what is reason for using TCP as the underlying transport protocol instead of the default? I've configured the link with a reasonably high cost in case I make another connection somewhere more local. >>>> This is my first experience with DECnet over more than a few local nodes, is there an easy way to get a full node list onto my systems? >>>> This all takes me back a while. Last commercial VMS/DECnet systems I used were back in 1994. I actually stumbled across hecnet whilst trying to find out if anyone still used uucp for anything. I was the last sysadmin of the UKNET uucp setup, then managed by PSINet before we finally dismantled all that infrastructure in 1999. >>>> Douglas >>> >>> -- >>> Johnny Billquist || "I'm on a bus >>> || on a psychedelic trip >>> email: bqt at softjar.se || Reading murder books >>> pdp is alive! || tryin' to stay hip" - B. Idol > > -- > Johnny Billquist || "I'm on a bus > || on a psychedelic trip > email: bqt at softjar.se || Reading murder books > pdp is alive! || tryin' to stay hip" - B. Idol