Archive-Date: Fri, 2 Oct 1998 13:41:18 -0400 Message-ID: From: =?iso-8859-1?Q?Ramiro_J=E1tiva?= Reply-To: Info-TCPware@process.com To: "'Info-TCPware@process.com'" CC: =?iso-8859-1?Q?Ramiro_J=E1tiva?= Subject: RV: #16936 Date: Fri, 2 Oct 1998 12:40:44 -0500 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable I have a question for the Newsgroup: Please respond your advice or comments to jativar@ccbvq.com Regards, Ramiro System Department in Chief Quito Stock Exchange > -----Mensaje original----- > De: Gerald Moran [SMTP:moran@process.com] > Enviado el: Viernes 2 de Octubre de 1998 09:56 AM > Para: 'Ramiro J=E1tiva' > Asunto: RE: #16936 >=20 > Hi Ramiro, >=20 > In the Technical Support Dept., we are not really equipped to handle > programming questions. > For such issues I would suggest that you submit these to the > Newsgroup: >=20 > > Info-TCPware@process.com=20 > >=20 > Developers do respond to postings that appear there. >=20 >=20 > Thank you. >=20 > Jerry Moran > Process Software Technical Support > 800-722-7770, x380 > moran@process.com >=20 >=20 > > ---------- > > From: Ramiro J=E1tiva[SMTP:jativar@ccbvq.com] > > Sent: Thursday, October 01, 1998 1:49 PM > > To: 'moran@process.com' > > Cc: Ramiro J=E1tiva > > Subject: #16936 > >=20 > > Dear Gerald: > >=20 > > Good news!. I followed your advice and wrote a server aplication > using > > the echo service of TCPware. The echo server at the Alpha side can > talk > > with the Client in Visual Basic using the socket Wrench. How ever = I > > need if you do not mind ask to you about certain details: > >=20 > > 1) The server program is write in DEC C and wait for client > connections. > > When the server is executed at the server side (Alpha server), the > > server aplication wait for connections as a process in the alpha > server. > >=20 > > 2) The client request a connection to the server, specific to the > echo > > server (@ip), the server create a connection through TCPWARE echo > server > > and NETCU can display using the SHOW CONNECTIONS command the = status > of > > the connections as: > >=20 > > ID RECVQ SENDQ Local Address Foreign Address > > State > > TCP86 0 0 132.1.1.2 > 132.1.1.13.1248 > > ESTABLISHED > >=20 > > saying that the client 132.1.1.13 is connected and ready to = transmit > > message to the server using the 1248 network port asigned by > TCPWARE. > > Additional the status displayed of the connection said that the > > connection was establiseh with the server 132.1.1.2, and all is OK. > >=20 > > My question is why exist situations where the server respond to the > > client connection and other cases that the server do not respond to > the > > client request. In other words, I can see that NETCU register all > > request of connections from the client, how ever the server > applications > > do not register the request of connection and do not respond. > >=20 > > I think that maybe the port assigned by TCPWARE to the connection > > request from the client do not fall in the same network port > assigned by > > Server Side when the client request the connection. In other > > situations, when the server attent the connection request from the > > client and accept the connection the client and the server can talk > at > > the same port network that is for example 1084. > >=20 > > In resume, I would like to know that maybe the solution is create a > > service call SERV1, that NETCU register as > >=20 > > ID Local Address > > SERV1 *.1084 > >=20 > > and said to my server aplication and my client aplication that can > talk > > using this network channel. Is rigth? > >=20 > > May be this can solve this problem. Could you advice to me if is > rigth? > >=20 > > > = ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > ++ > > ++++ > >=20 > > I was at the TCPWARE Web site, and I find the DECUS Fall '98. How > ever > > maybe I can not attent in this oportunity to the TCPWARE technical > > sessions in DECUS '98. I would like know information about TCP/IP > > Programming. Or maybe if you do not mind please send to me > information > > about other technical presentations of TCPWARE or other services as > > support/training. > > Thanks a lot for you patient. > >=20 > > With best regards, > > Ramiro > > Quito Stock Exchange > >=20 > >=20 > >=20 > >=20 ================================================================================ Archive-Date: Fri, 2 Oct 1998 13:52:17 -0400 Sender: bryant@process.com Date: Fri, 2 Oct 1998 13:48:59 -0400 From: Geoff Bryant Reply-To: Info-TCPware@process.com To: Info-TCPware@process.com CC: jativar@ccbvq.com, bryant@process.com Message-ID: <009CD193.DEA496D4.8@process.com> Subject: RE: RV: #16936 MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable It depends on how the server is written. =20 If the server is written to use the master server, then the master server c= an keep firing off server processes as client connect requests come in. If the server is written to handle accepting client connections itself, the= n it can be written to handle only one connection at a time, or to asynchronousl= y handle multiple clients. BTW, this is all pretty standard TCP/IP programming, so any good book may b= e handy, such as Rich Stevens' books or Doug Comer's books. Also, I can't = get to it now, but if send me an email, I can send you a document on some of th= e API issues for this with TCPware. It may take me a bit as I am out of the office for a week. =3D?iso-8859-1?Q?Ramiro_J=3DE1tiva?=3D writes: > >I have a question for the Newsgroup: > >Please respond your advice or comments to jativar=40ccbvq.com > >Regards, >Ramiro >System Department in Chief >Quito Stock Exchange > > >> -----Mensaje original----- >> De: Gerald Moran =5BSMTP:moran=40process.com=5D >> Enviado el: Viernes 2 de Octubre de 1998 09:56 AM >> Para: 'Ramiro J=E1tiva' >> Asunto: RE: #16936 >>=20 >> Hi Ramiro, >>=20 >> In the Technical Support Dept., we are not really equipped to handle >> programming questions. >> For such issues I would suggest that you submit these to the >> Newsgroup: >>=20 >> > Info-TCPware=40process.com=20 >> >=20 >> Developers do respond to postings that appear there. >>=20 >>=20 >> Thank you. >>=20 >> Jerry Moran >> Process Software Technical Support >> 800-722-7770, x380 >> moran=40process.com >>=20 >>=20 >> > ---------- >> > From: Ramiro J=E1tiva=5BSMTP:jativar=40ccbvq.com=5D >> > Sent: Thursday, October 01, 1998 1:49 PM >> > To: 'moran=40process.com' >> > Cc: Ramiro J=E1tiva >> > Subject: #16936 >> >=20 >> > Dear Gerald: >> >=20 >> > Good news=21. I followed your advice and wrote a server aplication >> using >> > the echo service of TCPware. The echo server at the Alpha side can >> talk >> > with the Client in Visual Basic using the socket Wrench. How ever I >> > need if you do not mind ask to you about certain details: >> >=20 >> > 1) The server program is write in DEC C and wait for client >> connections. >> > When the server is executed at the server side (Alpha server), the >> > server aplication wait for connections as a process in the alpha >> server. >> >=20 >> > 2) The client request a connection to the server, specific to the >> echo >> > server (=40ip), the server create a connection through TCPWARE echo >> server >> > and NETCU can display using the SHOW CONNECTIONS command the status >> of >> > the connections as: >> >=20 >> > ID RECVQ SENDQ Local Address Foreign Address >> > State >> > TCP86 0 0 132.1.1.2 >> 132.1.1.13.1248 >> > ESTABLISHED >> >=20 >> > saying that the client 132.1.1.13 is connected and ready to transmit >> > message to the server using the 1248 network port asigned by >> TCPWARE. >> > Additional the status displayed of the connection said that the >> > connection was establiseh with the server 132.1.1.2, and all is OK. >> >=20 >> > My question is why exist situations where the server respond to the >> > client connection and other cases that the server do not respond to >> the >> > client request. In other words, I can see that NETCU register all >> > request of connections from the client, how ever the server >> applications >> > do not register the request of connection and do not respond. >> >=20 >> > I think that maybe the port assigned by TCPWARE to the connection >> > request from the client do not fall in the same network port >> assigned by >> > Server Side when the client request the connection. In other >> > situations, when the server attent the connection request from the >> > client and accept the connection the client and the server can talk >> at >> > the same port network that is for example 1084. >> >=20 >> > In resume, I would like to know that maybe the solution is create a >> > service call SERV1, that NETCU register as >> >=20 >> > ID Local Address >> > SERV1 *.1084 >> >=20 >> > and said to my server aplication and my client aplication that can >> talk >> > using this network channel. Is rigth? >> >=20 >> > May be this can solve this problem. Could you advice to me if is >> rigth? >> >=20 >> > >> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >> ++ >> > ++++ >> >=20 >> > I was at the TCPWARE Web site, and I find the DECUS Fall '98. How >> ever >> > maybe I can not attent in this oportunity to the TCPWARE technical >> > sessions in DECUS '98. I would like know information about TCP/IP >> > Programming. Or maybe if you do not mind please send to me >> information >> > about other technical presentations of TCPWARE or other services as >> > support/training. >> > Thanks a lot for you patient. >> >=20 >> > With best regards, >> > Ramiro >> > Quito Stock Exchange >> >=20 >> >=20 >> >=20 >> >=20 ------------------------------------------------------------- Geoff Bryant bryant=40process.com TCPware/Multinet Engineering =09 Process Software Corporation http://www.process.com/ 959 Concord St. Framingham, MA 01701 USA ================================================================================ Archive-Date: Sat, 3 Oct 1998 05:15:36 -0400 Date: Sat, 3 Oct 1998 10:02:21 +0100 From: Steve Wakelin Reply-To: Info-TCPware@process.com Subject: Load balancing with Pathworks V5 and NT To: info-tcpware Message-ID: <98Oct3.101650bst.17921@gate.bgep.co.uk> MIME-Version: 1.0 Content-Type: text/plain I am am currently running a cluster to provide file services using Pathworks V5.0. OpenVMS V7.1-1h1 Pathworks V5.0F TCPware V5.2-3 We are migrating our existing Windows V3.1 /Decnet clients to NT Workstation V4.0. What we like to do is remove the NETBEUI traffic from the network. In order to achieve this I need to be able to load balance between the two node over TCP/IP. Pathworks V5.0 load balancing is usaually achived using NETBEUI. The main problem seems to be with the NT client. I seem unable to set DNS to be the primary lookup which I require for TCPware load balancing. As WINS always returns the IP address of the first node in the list of a multi-honed entry this does not achieve what we want. Has anyone managed to load balance over TCP/IP in this or a similar configuration? Thanks /Steve ================================================================================ Archive-Date: Sat, 3 Oct 1998 14:18:17 -0400 Message-ID: <36167178.5391B9F3@SMTP.DeltaTel.RU> Date: Sat, 03 Oct 1998 21:48:24 +0300 From: "Ruslan R. Laishev" Reply-To: Info-TCPware@process.com MIME-Version: 1.0 Subject: Re: Load balancing with Pathworks V5 and NT Content-Type: text/plain; charset=koi8-r Content-Transfer-Encoding: 7bit To: Info-TCPware@PROCESS.COM Steve Wakelin wrote: > > I am am currently running a cluster to provide file services > using Pathworks V5.0. > > OpenVMS V7.1-1h1 > Pathworks V5.0F > TCPware V5.2-3 I use OpenVMS V7.1-1h1 > Pathworks V6.0 > TCPware V5.3-2 > > We are migrating our existing Windows V3.1 /Decnet clients > to NT Workstation V4.0. The same. > > What we like to do is remove the NETBEUI traffic from the > network. I use several protocols: NETBEUI for locals M$ Windoze NT WS, and TCP/IP for WS beyond routers. > The main problem seems to be with the NT client. I seem > unable to set DNS to be the primary lookup which I require > for TCPware load balancing. If you have at WNT WS list of DNS server WS will use first always, 2-nd & 3-th will use when first is not reachable. > > As WINS always returns the IP address of the first node in > the list of a multi-honed entry this does not achieve what > we want. I use only _DNS_ for WINS resolutions, no problem. > Thanks > > /Steve -- Be weel. +----------------------------------------------Pure personal oppinion+ OpenVMS System/Network HardWorker Cel: 7+ (901) 971-3222 RSA/FngrPnt:D1C7F4D1912325A81C122F468A83293A Fido: 2:5030/279 +http://www.levitte.org/~rlaishev/ ----------------------------------+ ================================================================================ Archive-Date: Sat, 3 Oct 1998 16:21:53 -0400 From: engelandhv@freemail.nl (Engeland H. van) Reply-To: Info-TCPware@process.com Subject: Re: Load balancing with Pathworks V5 and NT Date: 3 Oct 1998 20:10:34 GMT Message-ID: <6v60bq$1e4$1@newnews.nl.uu.net> MIME-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII To: Info-TCPware@PROCESS.COM I at Gastec have the same configuration, and i do have load balancing. This is done with the Netbios alias configurated withing PCSA. For IP i use the IP loadbalancing as discribed in the TCPWARE manager book. I think its called cluster Ip address. It works. But as i stated the clients, dos with pcsa 5/6 client and Windows NT 4 clients, do there connections with a netbios call over Decnet or IP. I only use the IP load balancing with telnet or other IP trafic, or calls behind routers. Its not Betbeui but Netbios that does the call. The first node in the cluster that has the highest idle %, look with $PWMON/client (or just $PWMON) gets the client. So start the $ADMIN (pcsa interface) goto Utilities and Pathworks Alias. Refere to this name in the cluster IP address in TCPWARE. In tcpware enter a new IP address with this Alias name, less confusing, So in the TCPWARE_ROOT:[TCPWARE.NAMED]NAMED.HOSTS enter the alias like this 'pcsa_alias_name' IN A 'ip_address_first_cluster_node' IN A 'ip_address_second_cluster_node' IN HINFO Bla bla cluster_alias Forget the named.rev file. Enter this 'pcsa_alias_name' in Wins and see the trick is done, boss happy You BIG rease, You make shit, BOSS not happy boss angry boss throw you out. So test is before live In article <98Oct3.101650bst.17921@gate.bgep.co.uk>, wakelin.s@bgep.co.uk says... > > I am am currently running a cluster to provide file services > using Pathworks V5.0. > > OpenVMS V7.1-1h1 > Pathworks V5.0F > TCPware V5.2-3 > > We are migrating our existing Windows V3.1 /Decnet clients > to NT Workstation V4.0. > > What we like to do is remove the NETBEUI traffic from the > network. > > In order to achieve this I need to be able to load balance > between the two node over TCP/IP. > Pathworks V5.0 load balancing is usaually achived using > NETBEUI. > > The main problem seems to be with the NT client. I seem > unable to set DNS to be the primary lookup which I require > for TCPware load balancing. > > As WINS always returns the IP address of the first node in > the list of a multi-honed entry this does not achieve what > we want. > > Has anyone managed to load balance over TCP/IP in this or a > similar configuration? > > Thanks > > /Steve -- _____________________________________________________________________________ H. van Engeland Network Administrator/automation Consultant ENGELANDHV@FREEMAIL.NL GASTEC NV Dutch centre of Gas Technology Visit our site www.gastec.nl or www.gastec.com ================================================================================ Archive-Date: Tue, 13 Oct 1998 09:12:00 -0400 From: Hunter Goatley Reply-To: Info-TCPware@process.com Subject: Re: Multinet FTP vs UCX's VmsPlusMode, (site +vms+) Date: Tue, 13 Oct 1998 08:03:40 -0500 Message-ID: <3623095C.673D5FC8@PROCESS.COM> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: Info-TCPware@PROCESS.COM Larry D Bohan, Jr wrote: > > How are folks in general dealing w/ this? > Some sites install MadGoat FTP so they can use STRU O VMS. ftp://ftp.madgoat.com/madgoat/mgftp.zip > Are there any plans in the future for Multinet FTP > to support a UCX host sending files in this mode ? > No, there are no plans to do that at this time. > i recall (vaguely) that TcpWare could handle > this, but i don't recall if this was "as-is", > and/or 'useable, but unsupported' > It's supported, but STRU VMS is the preferred method. It's a lot cleaner (IMO) and doesn't create the unnecessary (when going from VMS to VMS) second file. Hunter ------ Hunter Goatley, Process Software, http://www.process.com/ http://www2.wku.edu/hunter/ ================================================================================ Archive-Date: Tue, 13 Oct 1998 15:32:14 -0400 Message-ID: <19981013192528.12289.rocketmail@send1e.yahoomail.com> Date: Tue, 13 Oct 1998 12:25:28 -0700 (PDT) From: Chris Wolfe Reply-To: Info-TCPware@process.com Subject: IMAP4 question To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Hi all, We're running TCPware 5.3-2 for VMS. I've tried using the IMAP4 server for my email, and discovered on two different occasions that my pc email client would lose it's "synch" with the actual messages on the server, like showing the wrong message body with the right email subject line in the index. It would also at times time-out in connecting to the server, and then would reply with "no new messages" even after I had received new ones. It seemed that most problems happened after something like this time-out error occurred. I would really like to move our company to IMAP4 instead of POP3, but I don't think I can recommend it now, seeing what has happened so far with my own account. Do you have any advice about the IMAP4 server, or know of any problems associated with it that others have experienced? Are there any configuration changes I can make? Thanks! Chris Wolfe _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ================================================================================ Archive-Date: Tue, 13 Oct 1998 15:59:27 -0400 Sender: goathunter@goat.process.com Date: Tue, 13 Oct 1998 14:55:38 -0500 From: Hunter Goatley Reply-To: Info-TCPware@process.com To: Info-TCPware@process.com Message-ID: <009CDA42.00C6ECBA.8@goat.process.com> Subject: RE: IMAP4 question Chris Wolfe writes: > >We're running TCPware 5.3-2 for VMS. I've tried using the IMAP4 >server for my email, and discovered on two different occasions that my >pc email client would lose it's "synch" with the actual messages on >the server, like showing the wrong message body with the right email >subject line in the index. It would also at times time-out in >connecting to the server, and then would reply with "no new messages" >even after I had received new ones. It seemed that most problems >happened after something like this time-out error occurred. > >I would really like to move our company to IMAP4 instead of POP3, but >I don't think I can recommend it now, seeing what has happened so far >with my own account. Do you have any advice about the IMAP4 server, >or know of any problems associated with it that others have >experienced? Are there any configuration changes I can make? > First, there are a few known problems with the IMAP4 server that we're working on. Hopefully, we'll have a new ECO kit in the next week or so. Second, from my playing around, the problem with files getting out-of-synch seems to happen when another browser/reader is used to access the mail messages. I could reproduce this easily using Netscape on an NT system. If I download the info, then use VMS Mail to move things around, the NT system gets very confused real fast. The only way I've found so far to clean that up is to delete the information files that Netscape creates. I'll be looking into this more this week and next to see if there's anything else that we can do to help with this. Hunter ------ Hunter Goatley, Process Software, http://www.process.com/ http://www2.wku.edu/hunter/ ================================================================================ Archive-Date: Tue, 13 Oct 1998 18:09:50 -0400 From: "Adam A. Luy" Reply-To: Info-TCPware@process.com Subject: off topic VMS question - ghost processes Date: Tue, 13 Oct 1998 13:17:52 -0700 Message-ID: <700cik$3va$1@fir.prod.itd.earthlink.net> To: Info-TCPware@PROCESS.COM Hello all, I have an Alpha 8200 running OpenVMS 7.1 and MUMPS/sql with over a dozen WAN locations connected. Occasionally we get disconnected or ghost processes that I am unable to kill with the usual 'STOP PROC /ID=' or even MST "-k" ##### commands. Our software vendor says that it is not possible to remove a disconnected process if the MST command doesn't kill it; they say you have to do a reboot. "Yeah right, when do you suggest I take down all of our ATM machines?" This is what one of the ghost processes looks like in a listing of processes, it's the one with the asterisks: Process Devices KB Directory Routine CPU,Glob Pri UIC Location 3C060 _VTA9941: 74 _DSA1:[ARZ] %VR 643058,45080 4 200,1 7D2E _VTA282: 197_DSA1:[ARZ] %VR 238442,22394 4 200,1 3970B ****** 37204 _VTA296: 205_DSA1:[ARZ] TSOs 416070,39891 4 200,1 37940 _VTA9990: 205_DSA1:[ARZ] CSTHI 824012,76087 4 200,1 As you can see there is no device associated, no mem, no CPU, heap sizes, etc.... Any idea on the real way to purge these bad boys? Any help is GREATLY appreciated, they eat up my MUMPS licenses. Adam A. Luy yulamada@earthlink.net ================================================================================ Archive-Date: Wed, 14 Oct 1998 09:14:09 -0400 Message-ID: <19981014130953.4041.rocketmail@send1d.yahoomail.com> Date: Wed, 14 Oct 1998 06:09:53 -0700 (PDT) From: Chris Wolfe Reply-To: Info-TCPware@process.com Subject: RE: IMAP4 question To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In fact, that's how I've cleared up the 'no new mail' problem, by going into VMS mail, deleting the first new message, and exiting. Then the mail client retrieves okay. Thanks for looking into it! ---Hunter Goatley wrote: > > Chris Wolfe writes: > > > >We're running TCPware 5.3-2 for VMS. I've tried using the IMAP4 > >server for my email, and discovered on two different occasions that my > >pc email client would lose it's "synch" with the actual messages on > >the server, like showing the wrong message body with the right email > >subject line in the index. It would also at times time-out in > >connecting to the server, and then would reply with "no new messages" > >even after I had received new ones. It seemed that most problems > >happened after something like this time-out error occurred. > > > >I would really like to move our company to IMAP4 instead of POP3, but > >I don't think I can recommend it now, seeing what has happened so far > >with my own account. Do you have any advice about the IMAP4 server, > >or know of any problems associated with it that others have > >experienced? Are there any configuration changes I can make? > > > First, there are a few known problems with the IMAP4 server that we're > working on. Hopefully, we'll have a new ECO kit in the next week or so. > > Second, from my playing around, the problem with files getting > out-of-synch seems to happen when another browser/reader is used to > access the mail messages. I could reproduce this easily using > Netscape on an NT system. If I download the info, then use VMS Mail > to move things around, the NT system gets very confused real fast. > The only way I've found so far to clean that up is to delete the > information files that Netscape creates. > > I'll be looking into this more this week and next to see if there's > anything else that we can do to help with this. > > Hunter > ------ > Hunter Goatley, Process Software, http://www.process.com/ > http://www2.wku.edu/hunter/ > _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ================================================================================ Archive-Date: Wed, 14 Oct 1998 15:37:13 -0400 Sender: goatley@triton.process.com Return-Path: From: Paul Coviello Reply-To: Info-TCPware@process.com Subject: tcpware and hp printers Date: Wed, 14 Oct 1998 13:31:13 -0400 Message-ID: <3624DFE1.58F56953@lmco.com> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="------------3C3948DFB70B4746829556D2" To: Info-TCPware@PROCESS.COM --------------3C3948DFB70B4746829556D2 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Does anybody know what I need to do to print a B size postscript on an HP printer, I can do A size ps but not B. I have tried everything I could find in manuals. Thanks Paul Coviello Lockheed Martin EIS - Sanders 95 Canal St. PO Box 868 Nashua, N.H. 03061-0868 * Voice: 603-885-4541 Fax: 603-885-3850 * email: Paul.Coviello@lmco.com --------------3C3948DFB70B4746829556D2 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Does anybody know what I need to do to print a B size postscript on an HP printer,  I can do A size ps but not B.  I have tried everything I could find in manuals.

Thanks
Paul Coviello
Lockheed Martin EIS - Sanders
95 Canal St. PO Box 868
Nashua, N.H. 03061-0868

* Voice: 603-885-4541  Fax: 603-885-3850
* email: Paul.Coviello@lmco.com
  --------------3C3948DFB70B4746829556D2-- ================================================================================ Archive-Date: Fri, 16 Oct 1998 00:41:41 -0400 Subject: Re: off topic VMS question - ghost processes Message-ID: <1998Oct15.174940@alcor.process.com> From: VOLZ@PROCESS.COM (Bernie Volz) Reply-To: Info-TCPware@process.com Date: 15 Oct 98 17:49:40 -0400 To: Info-TCPware@PROCESS.COM In article <700cik$3va$1@fir.prod.itd.earthlink.net>, "Adam A. Luy" writes: > Hello all, > > I have an Alpha 8200 running OpenVMS 7.1 and MUMPS/sql with over a dozen WAN > locations connected. Occasionally we get disconnected or ghost processes > that I am unable to kill with the usual 'STOP PROC /ID=' or even MST "-k" > ##### commands. Our software vendor says that it is not possible to remove a > disconnected process if the MST command doesn't kill it; they say you have > to do a reboot. "Yeah right, when do you suggest I take down all of our ATM > machines?" > > Any idea on the real way to purge these bad boys? Any help is GREATLY > appreciated, they eat up my MUMPS licenses. What does SHOW SYSTEM show? What state are these processes in? What does SHOW PROCESS/ALL/ID= show on the process? If you go into SDA, is there still outstanding I/O on the process (SHOW PROCESS/ID=). Likely, the process is in a state where STOP/ID and other things can't kill it because it is deadlocked waiting for something. The only solution is to find out why it is deadlocked to see if that can be prevented in the first place. - Bernie Volz Process Software ================================================================================ Archive-Date: Fri, 16 Oct 1998 17:03:38 -0400 From: RSierrajr@aol.com Reply-To: Info-TCPware@process.com Message-ID: <9b0aa603.3627af37@aol.com> Date: Fri, 16 Oct 1998 16:40:23 EDT To: Info-TCPware@process.com MIME-Version: 1.0 Subject: Re: off topic VMS question - ghost processes Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit >> I have an Alpha 8200 running OpenVMS 7.1 and MUMPS/sql with over a dozen WAN >> locations connected. Occasionally we get disconnected or ghost processes >> that I am unable to kill with the usual 'STOP PROC /ID=' or even MST "-k" >> ##### commands. Our software vendor says that it is not possible to remove a >> disconnected process if the MST command doesn't kill it; they say you have >> to do a reboot. "Yeah right, when do you suggest I take down all of our ATM >> machines?" >> >> Any idea on the real way to purge these bad boys? Any help is GREATLY >> appreciated, they eat up my MUMPS licenses. > >What does SHOW SYSTEM show? What state are these processes in? What does >SHOW PROCESS/ALL/ID= show on the process? If you go into SDA, is there >still outstanding I/O on the process (SHOW PROCESS/ID=). > >Likely, the process is in a state where STOP/ID and other things can't >kill it because it is deadlocked waiting for something. The only solution >is to find out why it is deadlocked to see if that can be prevented in >the first place. > >- Bernie Volz > Process Software "Ghost" processes are usually defined as processes that are gone at the VMS level but still exist in mumps. If this is the case, then no VMS show commands are going to see it. If the process still exists at the VMS level but is "hung" at the mumps level, you are out of luck. It is my understanding that the vendor (I assume Intersystems) sets a bit that prevents the processes from being STOPped by users. That is why they said the only way to clear them is to reboot. Ralph Sierra ================================================================================ Archive-Date: Mon, 19 Oct 1998 16:54:57 -0400 Message-ID: From: =?iso-8859-1?Q?Ramiro_J=E1tiva?= Reply-To: Info-TCPware@process.com To: "'Info-TCPware@process.com'" CC: =?iso-8859-1?Q?Ramiro_J=E1tiva?= Subject: Network Channel Date: Mon, 19 Oct 1998 15:53:41 -0500 MIME-Version: 1.0 Content-Type: text/plain I have a question for the Newsgroup: I would like ask to you how can I solve the following situation. I wrote a UDP Server that works very well using the UDPDriver Services that TCPWARE has. How ever each time that I startup the UDP server program that I development; TCPWARE assign a new network channel, and this network channel each time that the server is startup has a new number. My question is : How can I say to my program (called ServerUDP.c) that use for example the same channel each time that the server is startup? For example I would like to use 1070 as my network Channel all the times. Thanks a lot for your help. With best regards, Ramiro Please respond your advice or comments to jativar@ccbvq.com With best Regards, Ramiro System Department in Chief Quito Stock Exchange ================================================================================ Archive-Date: Mon, 19 Oct 1998 16:59:20 -0400 Sender: bryant@process.com Date: Mon, 19 Oct 1998 16:55:17 -0400 From: Geoff Bryant Reply-To: Info-TCPware@process.com To: Info-TCPware@process.com CC: bryant@process.com Message-ID: <009CDF09.B6ABC00C.49@process.com> Subject: RE: Network Channel There is no user control over the device name. The UDP device is created by VMS when you assign your channel to the UDP0: device, and VMS clones a new device. ------------------------------------------------------------------------------- I have a question for the Newsgroup: I would like ask to you how can I solve the following situation. I wrote a UDP Server that works very well using the UDPDriver Services that TCPWARE has. How ever each time that I startup the UDP server program that I development; TCPWARE assign a new network channel, and this network channel each time that the server is startup has a new number. My question is : How can I say to my program (called ServerUDP.c) that use for example the same channel each time that the server is startup? For example I would like to use 1070 as my network Channel all the times. Thanks a lot for your help. With best regards, Ramiro Please respond your advice or comments to jativar@ccbvq.com With best Regards, Ramiro System Department in Chief Quito Stock Exchange ================================================================================ Archive-Date: Wed, 21 Oct 1998 14:46:45 -0400 Message-ID: <19981021184203.2953.rocketmail@send1b.yahoomail.com> Date: Wed, 21 Oct 1998 11:42:03 -0700 (PDT) From: Chris Wolfe Reply-To: Info-TCPware@process.com Subject: Getting the ".com" added to an email address To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii First of all, we had this problem a year or longer ago, and while I like to think I document everything, I can't find or remember how it got resolved! :^) Ah, systems work... We're running TCPware 5.3-2 on Ovms. We have a user sending mail to an address with a domain of mail.bhs.boonville.k12.mo.us. He gets it returned since TCPware SMTP appended a .com to the end of the domain. As a test, we CAN send mail to integrasoft.columbia.mo.us, which may be the one we had a problem with a long time ago. Can you please let me know what the fix is? Thanks! Chris W _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ================================================================================ Archive-Date: Wed, 21 Oct 1998 15:21:35 -0400 Sender: schreiber@process.com Date: Wed, 21 Oct 1998 15:17:25 -0400 From: Jeff Schreiber Reply-To: Info-TCPware@process.com To: Info-TCPware@process.com CC: schreiber@process.com Message-ID: <009CE08E.5F83469D.46@process.com> Subject: RE: Getting the ".com" added to an email address Chris Wolfe writes: > >First of all, we had this problem a year or longer ago, and while I >like to think I document everything, I can't find or remember how it >got resolved! :^) Ah, systems work... We're running TCPware 5.3-2 on >Ovms. > >We have a user sending mail to an address with a domain of >mail.bhs.boonville.k12.mo.us. He gets it returned since TCPware SMTP >appended a .com to the end of the domain. As a test, we CAN send mail >to integrasoft.columbia.mo.us, which may be the one we had a problem >with a long time ago. > >Can you please let me know what the fix is? Thanks! > What is the value of your TCPWARE_DOMAINNAME? the lookup order should be: MX lookup of specified name. MX lookup of specified name with domainname appended A lookup of specified name. A lookup of specified name with domainname appended. But the TCPWARE_DOMAINNAME logical is a little misleading, since it's actually the full system name. So, for example, if my TCPWARE_DOMAINNAME logical is "host.process.com" than I would first look up an MX record for mail.bhs.boonville.k12.mo.us (which doesn't exist) than before I look for an A record, I would look for mail.bhs.boonville.k12.mo.us.process.com If my TCPWARE_DOMAINNAME "process.com", I would look up the fully qualified first (which doesn't exist), than I would append the domain name [.com] and look for an MX record mail.bhs.boonville.k12.mo.us.com Which happens to exist: mail.bhs.boonville.k12.mo.us.com IN MX 100 hustle.rahul.net And because that's found, we never check for the A record for mail.bhs.boonville.k12.mo.us. -Jeff -- Jeff Schreiber, Process Software Corp. schreiber@mx.process.com http://www.process.com TCPware & MultiNet: Stronger than Ever ================================================================================ Archive-Date: Wed, 21 Oct 1998 16:27:48 -0400 Message-ID: <19981021202423.7868.rocketmail@send105.yahoomail.com> Date: Wed, 21 Oct 1998 13:24:23 -0700 (PDT) From: Chris Wolfe Reply-To: Info-TCPware@process.com Subject: RE: Getting the ".com" added to an email address To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii The TCPWARE_DOMAINNAME was "carfax.com", so the .com WAS getting tried, and since there was an MX record, it failed, like you showed. I changed the logical to "www.carfax.com" and got this back when sending email again: > Error sending mail to : > Server replied: "550 relaying to denied; see http://www.rahul.net/howto/mailerrors.html" On that web page, it says this: We have also had reports that some sites are using software with bugs that takes an address ending in xx.US and appends .COM to it, generating an incorrect address of the form xx.US.COM. The buggy software will then try to deliver it to our machines, which are MX hosts for *.US.COM, and will get an error message denying relaying. The sending site should correct its software. But if this were still happening, we'd get a message like the first one, b/c it'd be trying the *.com again? Thanks again! ------------------------ Jeff Schreiber wrote: > What is the value of your TCPWARE_DOMAINNAME? > > the lookup order should be: > > MX lookup of specified name. > MX lookup of specified name with domainname appended > A lookup of specified name. > A lookup of specified name with domainname appended. > > But the TCPWARE_DOMAINNAME logical is a little misleading, since > it's actually the full system name. > > So, for example, if my TCPWARE_DOMAINNAME logical is "host.process.com" > than I would first look up an MX record for > mail.bhs.boonville.k12.mo.us (which doesn't exist) than before I look > for an A record, I would look for mail.bhs.boonville.k12.mo.us.process.com > > If my TCPWARE_DOMAINNAME "process.com", I would look up the fully > qualified first (which doesn't exist), than I would append the domain > name [.com] and look for an MX record mail.bhs.boonville.k12.mo.us.com > > Which happens to exist: > > mail.bhs.boonville.k12.mo.us.com IN MX 100 hustle.rahul.net > > And because that's found, we never check for the A record for > mail.bhs.boonville.k12.mo.us. > > -Jeff > > -- > Jeff Schreiber, Process Software Corp. > schreiber@mx.process.com http://www.process.com > TCPware & MultiNet: Stronger than Ever > _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ================================================================================ Archive-Date: Thu, 22 Oct 1998 00:45:27 -0400 Subject: Re: Network Channel Message-ID: <1998Oct21.235616@alcor.process.com> From: VOLZ@PROCESS.COM (Bernie Volz) Reply-To: Info-TCPware@process.com Date: 21 Oct 98 23:56:16 -0400 To: Info-TCPware@PROCESS.COM In article , =?iso-8859-1?Q?Ramiro_J=E1tiva?= writes: > I have a question for the Newsgroup: > > I would like ask to you how can I solve the following > situation. I wrote a UDP Server that works very well using the > UDPDriver Services that TCPWARE has. How ever each time that I startup > the UDP server program that I development; TCPWARE assign a new network > channel, and this network channel each time that the server is startup > has a new number. My question is : How can I say to my program (called > ServerUDP.c) that use for example the same channel each time that the > server is startup? For example I would like to use 1070 as my network > Channel all the times. > > Thanks a lot for your help. > > With best regards, > Ramiro Can you be a bit more specific about what you mean by network channel? I assume you are not talking about the VMS I/O channel (returned by SYS$ASSIGN) or the UDP device's unit number, but instead the PORT NUMBER used in packets? As far as the port number goes, if you don't "bind" to a specific port number, TCPware will assign a number for you. Typically, these numbers start at 1024 when the system is booted and increase each time a new number is assigned. There are two port numbers involved - the source and destination port number. You can issue a IO$_SETMODE request to specify the source port number for datagrams you sent. Please see the IO$_SETMODE function in the Programmer's Guide for UDPDRIVER. - Bernie Volz Process Software ================================================================================ Archive-Date: Thu, 22 Oct 1998 15:02:36 -0400 Message-ID: <19981022185649.4535.rocketmail@send101.yahoomail.com> Date: Thu, 22 Oct 1998 11:56:49 -0700 (PDT) From: Chris Wolfe Reply-To: Info-TCPware@process.com Subject: POP3 mail retrieval To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii We're on a new Alpha box, running TCPware 5.3-2. We've noticed that POP3 mail retrieval is much slower than on our previous Alpha. What can I check? Thanks! Chris W _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ================================================================================ Archive-Date: Fri, 23 Oct 1998 00:22:54 -0400 Subject: Re: POP3 mail retrieval Message-ID: <1998Oct22.230512@alcor.process.com> From: VOLZ@PROCESS.COM (Bernie Volz) Reply-To: Info-TCPware@process.com Date: 22 Oct 98 23:05:12 -0400 To: Info-TCPware@PROCESS.COM In article <19981022185649.4535.rocketmail@send101.yahoomail.com>, Chris Wolfe writes: > We're on a new Alpha box, running TCPware 5.3-2. We've noticed that > POP3 mail retrieval is much slower than on our previous Alpha. What > can I check? > > Thanks! > What's slow? Does it take a long time to connect to the server or is it slow in retrieving individual messages? ================================================================================ Archive-Date: Fri, 23 Oct 1998 09:12:47 -0400 Message-ID: <19981023130758.19195.rocketmail@send1d.yahoomail.com> Date: Fri, 23 Oct 1998 06:07:58 -0700 (PDT) From: Chris Wolfe Reply-To: Info-TCPware@process.com Subject: Re: POP3 mail retrieval To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In retrieving messages. Connecting is fine. ---Bernie Volz wrote: > > In article <19981022185649.4535.rocketmail@send101.yahoomail.com>, Chris Wolfe writes: > > We're on a new Alpha box, running TCPware 5.3-2. We've noticed that > > POP3 mail retrieval is much slower than on our previous Alpha. What > > can I check? > > > > Thanks! > > > > What's slow? Does it take a long time to connect to the server or is it > slow in retrieving individual messages? > _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ================================================================================ Archive-Date: Fri, 23 Oct 1998 11:55:54 -0400 From: "Brian Steele" Reply-To: Info-TCPware@process.com Subject: Problems with A1Mail, TCPWare 5.2... Date: Fri, 23 Oct 1998 11:14:36 -0400 Message-ID: <70q6bm$r581@col3.caribsurf.com> To: Info-TCPware@PROCESS.COM Hi Everyone: We use A1Mail on our LAN as our e-mail platform. The clients are Win95 PC clients with PATHWORKS installed. In August 1997, we started switching from DECnet/LAT to TCP/IP. As part of this switchover, we configured the A1Mail server to use TCP/IP as well as DECnet. All of the clients were then configured to use TCP/IP for Teamlinks Mail. There are about 160 PC clients in total. Since then we've been having some problems, and they are growing more severe, to the point now that we are forced to reboot the VAX at least once a week and usually more to clear problems with A1Mail. The server process goes into MUTEX state, and and this state cannot be cleared without a reboot. You can imagine the problems this causes on a production system! Initially the VAX used to "hang" shortly after the mail server process went into MUTEX. It would stop responding to TELNET logins, then stop responding entirely. I discovered that it was possible to keep the VAX running (and perform a graceful shutdown) if I used NETCU to cancel all the IP links associated with A1Mail, then run the A1Mail shutdown processes (NOTE - this still left the A1Mail server process in MUTEX state, but all other processes associated with A1Mail shut down normally). I've been in touch with the support staff at Process Software about the problem. One of them logged on to our VAX, had a look at the system, then suggested that the problem lay with the AMail server, specifically the BTYLM limit for the owner of the process. However, we've increased the BYTLM for the A1Mail server process from 8192 all the way up to 500,000, and the problem still occurs. The problem seems to happen at random occasions, but also seems to occur within 24 hours of someone mailing a message to all other users using a distribution list. But this could just be coincidence though. Are any of you successfully using A1Mail with TCPWare (or any other IP stack?). Have you seen this problem before? Any ideas on what might be causing it? Thanks and regards, Brian Steele ================================================================================ Archive-Date: Fri, 23 Oct 1998 14:59:35 -0400 Date: Fri, 23 Oct 1998 19:46:23 +0100 From: Steve Wakelin Reply-To: Info-TCPware@process.com Subject: DNS and Dynamic load balancing To: info-tcpware CC: support Message-ID: <98Oct23.200026bst.17937@gate.bgep.co.uk> MIME-Version: 1.0 Content-Type: text/plain We are running TCPWare V5.2-3. I am using the extensions within TCPware to allow for cluster load balancing. All clusters will be defined in a sub-domain delegated from our root domain. When the IP address is returned the TTL is 15 secs. We will not be using TCPware for our primary DNS server for the root domain. As the TTL is 15 secs for that period the DNS lookup server caches the entry and effectivly round-robins the addresses irrespective of load. Is it possible to lower the TTL to 0 (zero) so that the returned address is not cached. Each subsequent request for the clusters will require a lookup on the TCPware node and true load balancing will occur? TIA /Steve ================================================================================ Archive-Date: Fri, 23 Oct 1998 18:12:17 -0400 Subject: Re: POP3 mail retrieval Message-ID: <1998Oct23.173231@alcor.process.com> From: VOLZ@PROCESS.COM (Bernie Volz) Reply-To: Info-TCPware@process.com Date: 23 Oct 98 17:32:31 -0400 To: Info-TCPware@PROCESS.COM > In retrieving messages. Connecting is fine. > > ---Bernie Volz wrote: >> >> In article <19981022185649.4535.rocketmail@send101.yahoomail.com>, > Chris Wolfe writes: >> > We're on a new Alpha box, running TCPware 5.3-2. We've noticed that >> > POP3 mail retrieval is much slower than on our previous Alpha. What >> > can I check? Humm ... that seems a bit odd. Is VMS MAIL access pretty normal in terms of performance? Perhaps doing a packet trace of the POP3 traffic might be useful? $ netcu debug/tcp/data=1500/lpn=110 However, with a lot of connections active it will be very confusing and the performance many be difficult to determine. How is the performance of other TCP/IP applications - such as FTP? How many POP3 connections do you typically have? $ netcu show conn/local=*.pop3 Can you do a SHOW PROCESS/ALL/ID= process. Is the priority of this process 4 or higher? - Bernie Volz Process Software ================================================================================ Archive-Date: Fri, 23 Oct 1998 19:41:12 -0400 Subject: Re: DNS and Dynamic load balancing Message-ID: <1998Oct23.172613@alcor.process.com> From: VOLZ@PROCESS.COM (Bernie Volz) Reply-To: Info-TCPware@process.com Date: 23 Oct 98 17:26:13 -0400 To: Info-TCPware@PROCESS.COM In article <98Oct23.200026bst.17937@gate.bgep.co.uk>, Steve Wakelin writes: > We are running TCPWare V5.2-3. > > I am using the extensions within TCPware to allow for > cluster load balancing. All clusters will be defined in a > sub-domain delegated from our root domain. > > When the IP address is returned the TTL is 15 secs. We will > not be using TCPware for our primary DNS server for the root > domain. > > As the TTL is 15 secs for that period the DNS lookup server > caches the entry and effectivly round-robins the addresses > irrespective of load. > > Is it possible to lower the TTL to 0 (zero) so that the > returned address is not cached. Each subsequent request for > the clusters will require a lookup on the TCPware node and > true load balancing will occur? > TIA > > /Steve Check what the "minimum" TTL is on the SOA record for the zone. Most DNS servers also enforce some lower minimum (15 seconds sounds like it might just be that fixed minimum). In that, there likely isn't that much you can do about it. - Bernie Volz Process Software ================================================================================ Archive-Date: Sun, 25 Oct 1998 10:44:41 -0400 From: royalef@aol.com (Royal E. Frazier Jr.) Subject: Re: Problems with A1Mail, TCPWare 5.2... Date: Sun, 25 Oct 1998 14:31:34 GMT Sender: royalef@aol.com (Royal E. Frazier Jr.) Message-ID: <36333489.2972355@nntp.ix.netcom.com> Reply-To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Keywords: GIF animations, genealogy To: Info-TCPware@PROCESS.COM Brian, we've had some problems with some of our vaxes/a1/ip/tcpware. They sound a little different but similar situation. The problem for us appears to be triggered by leaving a TCP/IP pc logged into teamlinks using IP over the weekend when the various housekeeping routines run. I'm not involved in the problem, butI know nobody can log in on monday using IP. It requires killing the Ip connections (which I believe read FIN-WAIT2 in NETCU) and shutting down all in 1. I believe both cabinet and aida servers. This has been a long standing problem and I'm unsure who is primarily at fault. However, it seems to be that no IP connection should be capable of hanging an all-in-1 service...period. Sounds like a major achilles heel. With Digitial developing a web interfaces to all-in-1 I'd think they'd need to make sure that a idle Teamlinks/IP session can't knock out the whole Ip/service. Tcpware always seems unaffected by the situation. All-in-1 3.1, and 3.2 TCPWARE 5.1 & 5.2 Vax VMS 6.3 and 7.1 "Brian Steele" wrote: >Hi Everyone: > >We use A1Mail on our LAN as our e-mail platform. The clients are Win95 PC >clients with PATHWORKS installed. > >In August 1997, we started switching from DECnet/LAT to TCP/IP. As part of >this switchover, we configured the A1Mail server to use TCP/IP as well as >DECnet. All of the clients were then configured to use TCP/IP for Teamlinks >Mail. There are about 160 PC clients in total. > >Since then we've been having some problems, and they are growing more >severe, to the point now that we are forced to reboot the VAX at least once >a week and usually more to clear problems with A1Mail. The server process >goes into MUTEX state, and and this state cannot be cleared without a >reboot. You can imagine the problems this causes on a production system! > >Initially the VAX used to "hang" shortly after the mail server process went >into MUTEX. It would stop responding to TELNET logins, then stop responding >entirely. I discovered that it was possible to keep the VAX running (and >perform a graceful shutdown) if I used NETCU to cancel all the IP links >associated with A1Mail, then run the A1Mail shutdown processes (NOTE - this >still left the A1Mail server process in MUTEX state, but all other processes >associated with A1Mail shut down normally). > >I've been in touch with the support staff at Process Software about the >problem. One of them logged on to our VAX, had a look at the system, then >suggested that the problem lay with the AMail server, specifically the BTYLM >limit for the owner of the process. However, we've increased the BYTLM for >the A1Mail server process from 8192 all the way up to 500,000, and the >problem still occurs. > >The problem seems to happen at random occasions, but also seems to occur >within 24 hours of someone mailing a message to all other users using a >distribution list. But this could just be coincidence though. > >Are any of you successfully using A1Mail with TCPWare (or any other IP >stack?). Have you seen this problem before? Any ideas on what might be >causing it? > > >Thanks and regards, >Brian Steele > > Royal Frazier ======================================== royalef@aol.com http://members.aol.com/royalef/royal.htm ======================================== Family Genealogy and GIF Animation ======================================== ================================================================================ Archive-Date: Mon, 26 Oct 1998 09:47:48 -0400 Sender: schreiber@process.com Date: Mon, 26 Oct 1998 08:43:28 -0400 From: Jeff Schreiber Reply-To: Info-TCPware@process.com To: Info-TCPware@process.com CC: schreiber@process.com Message-ID: <009CE445.2A74E7E4.4@process.com> Subject: RE: DNS and Dynamic load balancing Steve Wakelin writes: > > Is it possible to lower the TTL to 0 (zero) so that the > returned address is not cached. Each subsequent request for > the clusters will require a lookup on the TCPware node and > true load balancing will occur? > TIA > The TTL of 15 is just the default for the load balanced records, but you can override it by just setting the TTL on those A records. So if you want to specify 0 for the TTL, just put 0 for the TTL on those records. balanced.domain.com 0 IN A 10.0.0.1 0 IN A 10.0.0.2 0 IN A 10.0.0.3 However there are a couple of things you should know about. First, there is no guarantee that the primary server will honor the 0 TTL. There are a multitude of servers that allow the TTL minimum/maximum to be modified. In other words, instead of a minumum of 0 and a max of 1 week. In other words, if they setup a minimum ttl of 100 on the primary server, it'll cache it for 100 seconds. The other problem is that most BIND based DNS servers will _always_ cache, regardless of the TTL. What this means is that it takes the information it receives, the TTL, and the current time. It adds the TTL to the current time, calls that the expire time, and caches it. When it gets another query, it checks if the current time is greater than the expire time and if it is, the record is considered stale. However, if that second query comes in within the same second, then it won't notice that the other one is expired. So if the queries come in within a second of each other, it will still round robin them. -Jeff -- Jeff Schreiber, Process Software Corp. schreiber@mx.process.com http://www.process.com TCPware & MultiNet: Stronger than Ever ================================================================================ Archive-Date: Mon, 26 Oct 1998 11:54:24 -0400 Message-ID: <19981026154830.28242.rocketmail@send101.yahoomail.com> Date: Mon, 26 Oct 1998 07:48:30 -0800 (PST) From: Chris Wolfe Reply-To: Info-TCPware@process.com Subject: Re: POP3 mail retrieval To: Info-TCPware@process.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii > > In retrieving messages. Connecting is fine. > > > >> > We're on a new Alpha box, running TCPware 5.3-2. We've noticed that > >> > POP3 mail retrieval is much slower than on our previous Alpha. What > >> > can I check? > > Humm ... that seems a bit odd. Is VMS MAIL access pretty normal in terms of > performance? I don't think anyone is accessing their mail via VMS MAIL, just POP3, since alot of it is Internet email. > > Perhaps doing a packet trace of the POP3 traffic might be useful? > $ netcu debug/tcp/data=1500/lpn=110 > However, with a lot of connections active it will be very confusing and the > performance many be difficult to determine. Yeah, that would be A LOT of data to look at! There are three people that have reported this last week and one more today. I can attempt a trace of one or two of them, if that would help narrow this down. Would there be anything else on the system to watch during that time? > > How is the performance of other TCP/IP applications - such as FTP? I can pull a 5.5mb .zip file from this system to my desktop in around 6 seconds... This new server is also connected to a 10/100 switch running at 100mbps, whereas the old server was only 10mbps (I've read before that there could be problems with this, like collisions, etc? We'll look at that port.) FTPing the same file from the old system was also around 6 seconds. > > How many POP3 connections do you typically have? > $ netcu show conn/local=*.pop3 BG5470 0 0 172.16.20.8.pop3 172.16.50.151.1063 ESTABLISHED BG0 0 0 172.16.20.8.pop3 172.16.50.140.1153 TIMED-WAIT BG0 0 0 172.16.20.8.pop3 172.16.50.123.1169 TIMED-WAIT BG0 0 0 172.16.20.8.pop3 172.16.50.145.1389 TIMED-WAIT BG8 0 0 *.pop3 *.* LISTEN -- ----- ----- ------------- --------------- ----- BG5470 0 0 172.16.20.8.pop3 172.16.50.151.1063 ESTABLISHED BG5531 0 0 172.16.20.8.pop3 172.16.50.157.1299 ESTABLISHED BG5593 0 0 172.16.20.8.pop3 172.16.50.148.1271 ESTABLISHED BG5615 0 0 172.16.20.8.pop3 172.16.50.111.1155 ESTABLISHED BG0 0 0 172.16.20.8.pop3 172.16.50.146.1164 TIMED-WAIT BG8 0 0 *.pop3 *.* LISTEN So far this a.m. I've seen 8 at most, usually 4 or 3. What does the 'TIMED-WAIT' mean? I see alot of those. > Can you do a SHOW PROCESS/ALL/ID= process. Is the priority of > this process 4 or higher? Base priority is 4. > > - Bernie Volz > Process Software > Thanks! _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ================================================================================ Archive-Date: Tue, 27 Oct 1998 08:34:41 -0400 Sender: goatley@triton.process.com Return-Path: Subject: Re: POP3 mail retrieval Message-ID: <1998Oct26.203955@alcor.process.com> From: VOLZ@PROCESS.COM (Bernie Volz) Reply-To: Info-TCPware@process.com Date: 26 Oct 98 20:39:55 -0400 To: Info-TCPware@PROCESS.COM In article <19981026154830.28242.rocketmail@send101.yahoomail.com>, Chris Wolfe writes: > > So far this a.m. I've seen 8 at most, usually 4 or 3. What does the > 'TIMED-WAIT' mean? I see alot of those. > After a connection closes, it is placed in TIMED-WAIT to make sure that the last ACK got received by the peer. These connections exist in this state for about 1-2 minutes and this is very normal. - Bernie Volz Process Software ================================================================================ Archive-Date: Tue, 27 Oct 1998 12:08:58 -0400 From: "William Black" Reply-To: Info-TCPware@process.com To: Subject: TCPware 5.3-2. & FTP Date: Tue, 27 Oct 1998 16:40:32 -0000 Message-ID: <003301be01c8$838e9880$bb0dcac2@PCESS023.essential.co.uk> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit I think I have found a problem with TCPware 5.3-2. Running the following command $spawn/input=nl: ftp results in the following error %DCL-S-SPAWNED, process EAMON_1 spawned %DCL-S-ATTACHED, terminal now attached to process EAMON_1 %TCPWARE_FTP-F-ACCVIO, access violation, reason mask=00, virtual address=7FEB79E8, PC=7FF523FB, PSL=03C00000 %DCL-S-RETURNED, control returned to process Eamon I did not get this error on the same Vax using TCPware version 5.0-3. Also the spawned job works correctly if the /input=nl: parameter is left out. Tested on a Vax 4300 running OPENVMS 7.1. The application is using FTP from Digital Standard Mumps (DSM) on the Vax. The only way to run a DCL command file from DSM is to spawn a job, hence the need for a spawned process. The job is being initiated from a Unix host using a remote shell command and the default input device for this type of command on the Vax is a mailbox device which causes the FTP on the Vax to fail unless the input device is redefined as the null device (NL:). We have being using this method for about 6 years using TCPware and up until now it has worked successfully. We also have an Alpha running OPENVMS 7.1 and TCPware 5.3 and again it works fine on this configuration. I am FTPing a simple text file from the Vax to a Unix host (Digital Alpha). ================================================================================ Archive-Date: Tue, 27 Oct 1998 13:28:31 -0400 Sender: schreiber@process.com Date: Tue, 27 Oct 1998 13:28:20 -0400 From: Jeff Schreiber Reply-To: Info-TCPware@process.com To: Info-TCPware@process.com CC: schreiber@process.com Message-ID: <009CE536.208DE513.84@process.com> Subject: RE: TCPware 5.3-2. & FTP "William Black" writes: > >I think I have found a problem with TCPware 5.3-2. > > >The application is using FTP from Digital Standard Mumps (DSM) on the Vax. >The only way to run a DCL command file from DSM is to spawn a job, hence the >need for a spawned process. The job is being initiated from a Unix host >using a remote shell command and the default input device for this type of >command on the Vax is a mailbox device which causes the FTP on the Vax to >fail unless the input device is redefined as the null device (NL:). > It's definately a problem on a VAX (it seems to work on the alpha however). I would recommend you contact our support department in order to have the defect properly filed (with your contact information and all that). However, I'm not sure from your description how the commands are getting to FTP to copy the text file. There might be a way around what you are trying to do. Are you using a startup file to do the login and put, or is there something else that is missing? -Jeff -- Jeff Schreiber, Process Software Corp. schreiber@mx.process.com http://www.process.com TCPware & MultiNet: Stronger than Ever ================================================================================ Archive-Date: Tue, 27 Oct 1998 16:18:14 -0400 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: Info-TCPware@process.com From: "Daniel A Gauthier" Reply-To: Info-TCPware@process.com Subject: RE: TCPware 5.3-2. & FTP Date: Tue, 27 Oct 1998 16:17:48 -0500 Message-ID: <19981027211748578.AAA577.914@d175014.dynamic.cmich.edu> >>The application is using FTP from Digital Standard Mumps (DSM) on the Vax. >>The only way to run a DCL command file from DSM is to spawn a job, hence the >>need for a spawned process. The job is being initiated from a Unix host >>using a remote shell command and the default input device for this type of >>command on the Vax is a mailbox device which causes the FTP on the Vax to >>fail unless the input device is redefined as the null device (NL:). Did you try assigning the input to a blank file i.e. DUA0:[UTIL]BLANK.FIL unstead of the NL: ?? I recommend keeping a blank file around for just such occasions. (I also use it to: print/note="........." blank.fil ) Daniel A. Gauthier Telemanagement Analyst Central Michigan University Technology Operations Dept. Voice: (517) 774-1355 Fax: (517) 774-3537 Email: 3in7ifi@cmich.edu -------------------------------------------------------------------- My employer might hate everything I say here, so don't blame them. ================================================================================ Archive-Date: Tue, 27 Oct 1998 18:05:49 -0400 From: galen.tackett@lmco.com (Galen Tackett) Reply-To: Info-TCPware@process.com Subject: rsh not working on one of three clustered nodes Date: Tue, 27 Oct 1998 14:02:55 -0700 Message-ID: To: Info-TCPware@PROCESS.COM I have three nodes (call them A, B, and C) in a cluster that share a common system disk. TCPware V5.2-3 is running on all three. rsh commands into A or B work normally, but rsh commands into C just terminate (after a few minutes) with a "connection timed out" message. As you might expect, the same goes for rlogin attempts. This is true even of rsh/rlogin commands sent from C to itself. I compared the service definitions on all three nodes and they look the same. Next, I looked at NETCU SHOW CONNECTIONS on the originating node and on C. On the originating node, the connection just stays in the "SYN-SENT" state, while on C no connection ever shows up at all. I tried to use TCPDUMP to look at the data flow between C and the originating node, but for some reason TCPDUMP complains that my network interface (FWA0:, a DEFPA) is unknown. Does anyone have any idea what might be happening? -- Galen Tackett Lockheed Martin Missiles & Space *** Remove ".no.spam" from address before e-mailing me! *** ================================================================================ Archive-Date: Tue, 27 Oct 1998 18:35:58 -0400 Message-ID: <36365A52.A13CF06E@process.com> Date: Tue, 27 Oct 1998 18:42:10 -0500 From: Michael Corbett Reply-To: Info-TCPware@process.com MIME-Version: 1.0 To: info-tcpware@mx.process.com Subject: RE: rsh not working on one of three clustered nodes Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > I have three nodes (call them A, B, and C) in a cluster that share a > common system disk. TCPware V5.2-3 is running on all three. rsh commands > into A or B work normally, but rsh commands into C just terminate (after a > few minutes) with a "connection timed out" message. As you might expect, > the same goes for rlogin attempts. This is true even of rsh/rlogin > commands sent from C to itself. > > I compared the service definitions on all three nodes and they look the > same. Next, I looked at NETCU SHOW CONNECTIONS on the originating node and > on C. On the originating node, the connection just stays in the "SYN-SENT" > state, while on C no connection ever shows up at all. > > I tried to use TCPDUMP to look at the data flow between C and the > originating node, but for some reason TCPDUMP complains that my network > interface (FWA0:, a DEFPA) is unknown. > > Does anyone have any idea what might be happening? > > I'm not sure why you are having problems with RSH on the one node. The TCPDUMP might help to explain it. Try adding a /INTERFACE=IP to the TCPDUMP command do get around the errors. You might also want to get on the node that is having problems and try an RSH to the loopback (127.0.0.1) and see if that works. regards Mike -- +-------------------------------------------------------------------------+ Michael Corbett Email: Corbett@process.com Process Software Corporation Phone: 800 722-7770 x369 959 Concord St. 508 879-6994 x369 Framingham MA 01701-4682 FAX: 508 879-0042 ================================================================================ Archive-Date: Wed, 28 Oct 1998 13:43:21 -0400 From: "Mark Levy" Reply-To: Info-TCPware@process.com Subject: Multihoming, DNS, & NAT Date: Wed, 28 Oct 1998 12:36:35 -0600 Message-ID: <36376471.0@news.dls.net> To: Info-TCPware@PROCESS.COM I have several customers with systems behind firewalls or routers which do network address translation. Consequently, the systems which are accessible from "outside" (web servers, mail servers, etc.) have internal addresses and static translations for their globally-unique addresses. In DNS, I specify both addresses. Occasionally, the wrong address is given for a request; for example, the internal address will appear first in the list in an inquiry from "outside." This obviously delays access to the system while the first address times out and the second is tried. I've looked at the sortlist directive, but it appears that it will give preference to the internal or external address, but not both, depending on where the requester is located. Obviously, I'd like the internal address given preference for internal users, and the external likewise. Short of setting up two nameservers, is there a way to accomplish this? Thanks in advance. Mark Levy FSINet ================================================================================ Archive-Date: Wed, 28 Oct 1998 14:09:59 -0400 From: galen.tackett@lmco.com (Galen Tackett) Reply-To: Info-TCPware@process.com Subject: Re: rsh not working on one of three clustered nodes Date: Wed, 28 Oct 1998 09:45:55 -0800 Message-ID: To: Info-TCPware@PROCESS.COM In article <36365A52.A13CF06E@process.com>, Michael Corbett wrote: > > I'm not sure why you are having problems with RSH on the one node. > The TCPDUMP might help to explain it. Try adding a /INTERFACE=IP to the > TCPDUMP command do get around the errors. You might also want to get on the > node that is having problems and try an RSH to the loopback (127.0.0.1) and > see if that works. > Well, it turns out that this was due to a bone-headed edit of the DNS database that resulted in the system I called "C" having two addresses, only one of which was actually valid. Once I found this and cleared it up, rsh began working just fine. Before, it was trying to talk to the bogus address. -- Galen Tackett Lockheed Martin Missiles & Space *** Remove ".no.spam" from address before e-mailing me! ***