INFO-VAX Wed, 10 Jan 2007 Volume 2007 : Issue 20 Contents: Re: 1,000,000 License PAK's Served! Re: 1,000,000 License PAK's Served! Re: 1,000,000 License PAK's Served! Re: 1,000,000 License PAK's Served! Re: 1,000,000 License PAK's Served! Re: 1,000,000 License PAK's Served! Re: Am I a boot node ? Re: Cluster connection lost when one link fails? Finding textual information about a device and device driver INIT volume /VOLUME_CHAR=NOACCESS_DATES fails. Re: KSC CAMAC/VXI OpenVMS drivers Public Release Managing VMS$OBJECTS.DAT Re: Managing VMS$OBJECTS.DAT Re: MONITOR X-axis change ODS5 and hardlinks Re: ODS5 and hardlinks Re: ODS5 and hardlinks Re: ODS5 and hardlinks Re: ODS5 and hardlinks Re: On-shoring of VMS support?? Re: On-shoring of VMS support?? Star Trek and $CONFIGURE GALAXY Re: Star Trek and $CONFIGURE GALAXY Re: Star Trek and $CONFIGURE GALAXY Re: Star Trek and $CONFIGURE GALAXY Strategy for common DECNET proxy files in cluster ? Re: Stuck TCPIP connections Re: Stuck TCPIP connections Re: VMS732_UPDATE-V0900 fails ---------------------------------------------------------------------- Date: Wed, 10 Jan 2007 18:07:02 +1030 From: Mark Daniel Subject: Re: 1,000,000 License PAK's Served! Message-ID: <12q960nqibeh13b@corp.supernews.com> davidc@montagar.com wrote: > The OpenVMS Hobbyist Program has served it's one millionth License PAK! > > http://www.openvmshobbyist.com/blog/index.php Many thanks to David C., David T. of IC, and congratulations to David B.^h^h^h^h^h^h^h^hGeorg B. Is it possible to have more of a breakdown ... How many PAKs are issued per hobbiest? How many issued per hobbiest per year? How many years? (My query; it's a different milestone if 3,000 PAKs have been issued to 60 hobbiests every year for the last 6 or so.) ------------------------------ Date: Wed, 10 Jan 2007 01:58:08 -0600 (CST) From: sms@antinode.org (Steven M. Schweda) Subject: Re: 1,000,000 License PAK's Served! Message-ID: <07011001580829_2020028F@antinode.org> From: Mark Daniel > Is it possible to have more of a breakdown ... > > How many PAKs are issued per hobbiest? > [...] Better yet, can you break it down into "hobby", "hobbier", and "hobbiest"? (Or are they all actually "hobbyist"? AIEEE!!!) ------------------------------------------------------------------------ Steven M. Schweda sms@antinode-org 382 South Warwick Street (+1) 651-699-9818 Saint Paul MN 55105-2547 ------------------------------ Date: Wed, 10 Jan 2007 18:47:06 +1030 From: Mark Daniel Subject: Re: 1,000,000 License PAK's Served! Message-ID: <12q98bs5gqi91dd@corp.supernews.com> Steven M. Schweda wrote: > From: Mark Daniel > >>Is it possible to have more of a breakdown ... >> >>How many PAKs are issued per hobbiest? >>[...] > > > Better yet, can you break it down into "hobby", "hobbier", and > "hobbiest"? (Or are they all actually "hobbyist"? AIEEE!!!) > > ------------------------------------------------------------------------ > > Steven M. Schweda sms@antinode-org > 382 South Warwick Street (+1) 651-699-9818 > Saint Paul MN 55105-2547 Ok. Question withdrawern. I'm not sensitive. Doesn't concern me about YAHmail, YahMail, soyMail, Daniels, Wasd or San Paulo. Sheesh. Must be nice to have a spell-checker (or a liberal education). No bugger it. I'm still curious. David C., any further breakdown on the PAK statistics? Just curious about the hobbiest population. ------------------------------ Date: Wed, 10 Jan 2007 14:19:32 +0100 From: Paul Sture Subject: Re: 1,000,000 License PAK's Served! Message-ID: In article <12q9gh7iajff2b6@corp.supernews.com>, Mark Daniel wrote: > Paul Sture wrote: > > In article <12q960nqibeh13b@corp.supernews.com>, > > Mark Daniel wrote: > > > > > >>davidc@montagar.com wrote: > >> > >>>The OpenVMS Hobbyist Program has served it's one millionth License PAK! > >>> > >>>http://www.openvmshobbyist.com/blog/index.php > >> > >>Many thanks to David C., David T. of IC, and congratulations to David > >>B.^h^h^h^h^h^h^h^hGeorg B. > >> > >>Is it possible to have more of a breakdown ... > >> > >>How many PAKs are issued per hobbiest? > >> > >>How many issued per hobbiest per year? > >> > >>How many years? > >> > >>(My query; it's a different milestone if 3,000 PAKs have been issued to > >>60 hobbiests every year for the last 6 or so.) > > > > > > Looking at the Hobbyist licenses I renewed last week: > > No Paul; apparently I was looking for hobbiest licenses. How many of those? Don't let Steven Schweda's comment get to you. He can be a bit pedantic about grammar and spelling at times - perhaps he should add the occasional smiley. > > 1 VAX-VMS license (in my case for SIMH) > > 1 OPENVMS-ALPHA license > > 107 layered product licenses (ISTR it was 104 in previous years) > > > > The VMS licenses are NO_SHARE, so you need one per system for clustering > > purposes. One set of layered product licenses works for all the systems > > a Hobbyist has. > > > > 110 licenses per Hobbyist per year seems a reasonable ball park figure. > > I believe I got my first set in 1999. > > Thanks. Circa 9,000 hobbyists then? Seems a lot all the same. Divided > by how many years I wonder? http://www.openvmshobbyist.com/openvms_history/hobbyist_history/index.htm l "In May 1997 at the DECUS Symposia in Cincinnati The OpenVMS Hobbyist program began with a simple premise: Someone wanted to run the OpenVMS computer Operating System, not for money not for profit, but for the simple pleasure of using one of the finest Operating systems ever developed for their own personal use." So that's just over 9 years. However. even second hand kit was relatively expensive back then - my Islandco Alpha cost 2500 USD plus shipping and import duty in NOV-2001 (when the USD was high - think more like 5000 AUD). Yikes - at today's prices that could be an Impressive Cluster! Comparing my license numbers, approximately 70,000 PAKs have been issued in the last 12 months. -- Paul Sture ------------------------------ Date: 10 Jan 2007 08:09:40 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: 1,000,000 License PAK's Served! Message-ID: In article <07011001580829_2020028F@antinode.org>, sms@antinode.org (Steven M. Schweda) writes: > From: Mark Daniel > >> Is it possible to have more of a breakdown ... >> >> How many PAKs are issued per hobbiest? >> [...] > > Better yet, can you break it down into "hobby", "hobbier", and > "hobbiest"? (Or are they all actually "hobbyist"? AIEEE!!!) I'll be happy as long as the volunteers just keep the system running. I don't care if they keep statistics. ------------------------------ Date: Thu, 11 Jan 2007 00:02:48 +0800 From: prep@k9.prep.synonet.com Subject: Re: 1,000,000 License PAK's Served! Message-ID: <87k5zug993.fsf@k9.prep.synonet.com> JF Mezei writes: > Mark Daniel wrote: >> No Paul; apparently I was looking for hobbiest licenses. How many >> of those? > How do you define the bobbiest ? No, must not give in... ------------------------------ Date: Wed, 10 Jan 2007 18:19:40 +0000 (UTC) From: moroney@world.std.spaamtrap.com (Michael Moroney) Subject: Re: Am I a boot node ? Message-ID: JF Mezei writes: >Does a node know that it is acting as a boot node for another node ? Once >the initial boot file has been served via MOP by LANACP or DECNET, is >there any trace that the satellite node is depending on that node ? There is no dependence, as long as the satellite has access to its system disk from somewhere. It's theoretically possible, although I don't know of anyone ever to have done it, to boot a satellite from a node in another cluster, a nonclustered system or even a non-VMS node. All the node needs are a few certain files with enough knowledge to find and use its MSCP-served system disk. >Wouldn't the satellite then access the boot node's disk via MSCP just like >any other disk ? Exactly. >Or does the clustering protocol/software allow a boot node to know if >another node on the network depends on any of its disks for system software ? It could probably figure out if a system depends on a disk it has, if the disk isn't multiported somehow, or if it is and it's not seen by the MSCP alternate path, and it's not a MSCP-server-only device, another system could depend on that node for the disk. ------------------------------ Date: 10 Jan 2007 04:34:54 -0600 From: omond@encompasserve.org (Roy Omond) Subject: Re: Cluster connection lost when one link fails? Message-ID: In article <45a420a8$1@flight>, "Malcolm Dunnett" writes: > "Malcolm Dunnett" wrote in message > news:KAylh.106636$rv4.80875@edtnps90... >> VMS 7.3-2. Although since the problem is in the DE500 driver I wouldn't be >> surprised if the same problem occurs in later versions as well ( I suspect >> there's not been much work done on the DE500 driver in a while ) >> > > HP has confirmed that the fix for this problem will be included > in the VMS732_LAN-V0500 patch kit (not yet released). I'm a bit confused/worried. Is the patch addressing the issue *solely* for the DE500 ? I've observed this problem with DEGXAs, so it would appear to be not restricted to DE500s. Maybe it's generic in all Ethernet drivers ? Anyone like to comment ? ------------------------------ Date: Wed, 10 Jan 2007 08:58:33 -0500 From: JF Mezei Subject: Finding textual information about a device and device driver Message-ID: <45a4f213$0$28590$c3e8da3@news.astraweb.com> How does one go about to find out what a particular device and/or device driver is meant to be doing ? For instance, I take a random device on my system SMA0. It is SYS$SMDRIVER.EXE Looking at the contents of the executable there isn't any description/hint of what it does. Would I have to buy the sources just to find this out ? Or is there some file I can search on the system disk that contains a brief description of each driver that is loaded ? Some drivers are obvious. But others are not. ------------------------------ Date: Wed, 10 Jan 2007 07:55:19 -0500 From: JF Mezei Subject: INIT volume /VOLUME_CHAR=NOACCESS_DATES fails. Message-ID: <45a4e3c6$0$25340$c3e8da3@news.astraweb.com> Alpha VMS 8.3 HELP INIT/VOLUME INITIALIZE /VOLUME_CHARACTERISTICS /VOLUME_CHARACTERISTICS=([[NO]HARDLINKS,] [[NO]ACCESS_ DATES[=delta-time]]) <...> Note that the NOACCESS_DATES option affects only the node on which the command is issued. Other nodes are not affected by the change until the next time the volume is mounted. See the Guide to OpenVMS File Applications for additional information. 1- The above mentioned manual doesn't contain info on initialising ODS5 disks. 2- The following procedure fails: $INIT $11$dqb0: MAVIC - /CLUSTER=6 - /DIRECTORIES=100 - /EXTENSION=18 - /NOGPT - /HEADERS=45000 - ! approximate max number of files expected /HOMEBLOCKS=FIXED - /INDEX=MIDDLE - /OWNER=SYSTEM - /STRUCTURE=5 - /SYSTEM - /VOLUME_CHARACTERISTICS=(hardlinks,noaccess_dates) - /WINDOWS=48 %DCL-W-NOTNEG, qualifier or keyword not negatable - remove "NO" or omit \NOACCESS_DATES\ I assume that this documentation is shared with SET VOLUME for some items. But the DCL declaration for INIT does not allow NOACCESS_DATES to be specified. Is this a big or did I read the doc wrong ? If I initialise it without access_dates specified, then mount it, I can then do a SET VOLUME xxxx /NOACCESS_DATES with success. ------------------------------ Date: 10 Jan 2007 02:28:54 -0800 From: "Ian Miller" Subject: Re: KSC CAMAC/VXI OpenVMS drivers Public Release Message-ID: <1168424933.993869.43870@i56g2000hsf.googlegroups.com> I looked at the ftp site but could not see ksc09jan07Vms.zip ------------------------------ Date: Wed, 10 Jan 2007 04:44:13 -0500 From: JF Mezei Subject: Managing VMS$OBJECTS.DAT Message-ID: <45a4b0e8$0$30646$c3e8da3@news.astraweb.com> Looking at which files need to be common on all my systems, I noticed that VMS$OBJECTS.DAT contains references to old disks that will not exist anymore, as well as disk volumes from CDs that were mounted. Which command would be used to remove entries from this file ? ------------------------------ Date: 10 Jan 2007 02:09:12 -0800 From: "Ian Miller" Subject: Re: Managing VMS$OBJECTS.DAT Message-ID: <1168423752.645533.216140@p59g2000hsd.googlegroups.com> I think SET SECURITY can manage entries in this file. ------------------------------ Date: 10 Jan 2007 12:51:40 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: MONITOR X-axis change Message-ID: In article <1168449662.463864@proxy.dienste.wien.at>, "Ferry Bolhar" writes: > > Is there a way to increase the size of the X axis so that it can > reflect approximately the actual amount of processes? > I think this has been asked before and I think the answer is still no. I wonder if the original DISPLAY.EXE program from VMS 2 could be made open source? IIRC MONITOR is largely written in PL/I. Asking for enhancements in it from a company that doesn't even have a VMS PL/I compiler may run into a Catch-22 (although I assume they have a licensed copy from Kednos to use every time they rebuild VMS from source). You may be able to find a port of a popular utility, like "top" to VMS. I can recall trying to add some MONITOR like capabilities to top for HP-UX, but the C compiler that shipped with HP-UX was too painfull to bother with. At the time a K&R compiler shipped with HP-UX and an ANSI compiler was an extra cost option. K&R were not responsible for the problems I had with that compiler. ------------------------------ Date: Wed, 10 Jan 2007 08:18:58 -0500 From: JF Mezei Subject: ODS5 and hardlinks Message-ID: <45a4e75e$0$14718$c3e8da3@news.astraweb.com> That newfangled ODS5 beast is all foreign to me... I don't yet have 20 years experience with VMS (this comes in may this year :-), so pardon my newbieness in this matter... I've tried to find documentation that *really* described what hardlinks are all about. Best I could come up with is in comp.os.vms. Can someone confirm that SET VOLUME/VOLUME_CHARACTERISTICS=hardlinks (and the equivalent INIT /VOLUME_CHAR=HARDLINKS changes nothing in usage for the disk (AKA SET FILE/ENTER SET FILE/REMOVE ) except that it implements some form of usage count for each file so that a DELETE won't actually release the blocks unless there are no aliases left for the file ? Is this the only difference ? If one does a SET VOLUME on an existing disk, must one do a ANA/DISK/REPAIR to build those number of links to each file ? ------------------------------ Date: 10 Jan 2007 05:31:19 -0800 From: "AEF" Subject: Re: ODS5 and hardlinks Message-ID: <1168435878.532710.272420@77g2000hsv.googlegroups.com> JF Mezei wrote: > That newfangled ODS5 beast is all foreign to me... I don't yet have 20 > years experience with VMS (this comes in may this year :-), so pardon my > newbieness in this matter... > > I've tried to find documentation that *really* described what hardlinks are > all about. Best I could come up with is in comp.os.vms. > > Can someone confirm that SET VOLUME/VOLUME_CHARACTERISTICS=hardlinks (and > the equivalent INIT /VOLUME_CHAR=HARDLINKS changes nothing in usage for the > disk (AKA SET FILE/ENTER SET FILE/REMOVE ) except that it implements some > form of usage count for each file so that a DELETE won't actually release > the blocks unless there are no aliases left for the file ? > > Is this the only difference ? > > If one does a SET VOLUME on an existing disk, must one do a ANA/DISK/REPAIR > to build those number of links to each file ? Did you check http://groups.google.com/group/comp.os.vms/browse_frm/thread/80baa65b0a17a992/4a96c1f24e02d423?lnk=gst&q=HARD+LINKS+AEF&rnum=1&hl=en#4a96c1f24e02d423 ? AEF ------------------------------ Date: Wed, 10 Jan 2007 08:45:02 -0500 From: JF Mezei Subject: Re: ODS5 and hardlinks Message-ID: <45a4ed7b$0$14671$c3e8da3@news.astraweb.com> AEF wrote: > Did you check > > http://groups.google.com/group/comp.os.vms/browse_frm/thread/80baa65b0a17a992/4a96c1f24e02d423?lnk=gst&q=HARD+LINKS+AEF&rnum=1&hl=en#4a96c1f24e02d423 Yes. That thread was about alias files with just a brief mention of hardlinks. I just need some confirmation that there are no downsides to enabling hardlinks and to be aware of what impact it really has. ------------------------------ Date: Wed, 10 Jan 2007 06:37:51 -0800 From: "Tom Linden" Subject: Re: ODS5 and hardlinks Message-ID: On Wed, 10 Jan 2007 05:18:58 -0800, JF Mezei = wrote: > That newfangled ODS5 beast is all foreign to me... I don't yet have 20= = > years experience with VMS (this comes in may this year :-), so pardon = my = > newbieness in this matter... > > I've tried to find documentation that *really* described what hardlink= s = > are all about. Best I could come up with is in comp.os.vms. > > Can someone confirm that SET VOLUME/VOLUME_CHARACTERISTICS=3Dhardlinks= = > (and the equivalent INIT /VOLUME_CHAR=3DHARDLINKS changes nothing in u= sage = > for the disk (AKA SET FILE/ENTER SET FILE/REMOVE ) except that it = > implements some form of usage count for each file so that a DELETE won= 't = > actually release the blocks unless there are no aliases left for the = > file ? > > Is this the only difference ? > > If one does a SET VOLUME on an existing disk, must one do a = > ANA/DISK/REPAIR to build those number of links to each file ? I always found symbolic links more useful than hardlinks on Unix systems= . -- = Using Opera's revolutionary e-mail client: http://www.opera.com/mail/ ------------------------------ Date: 10 Jan 2007 12:31:40 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: ODS5 and hardlinks Message-ID: In article <45a4ed7b$0$14671$c3e8da3@news.astraweb.com>, JF Mezei writes: > > Yes. That thread was about alias files with just a brief mention of > hardlinks. I just need some confirmation that there are no downsides to > enabling hardlinks and to be aware of what impact it really has. I think the biggest problem is the same one you can find on UNIX: if you delete a file and others have hardlinks to it, what becomes of the disk quota? On UNIX I'm told the original owner is charged quota until the last link is removed. Pity the user who needs to recover some disk quota Right Now. I don't use disk quotas on any system I manage, so I don't know, but I won't enable hard links on VMS unless I prove to myself that I need them. ------------------------------ Date: 10 Jan 2007 02:41:08 -0800 From: "Ian Miller" Subject: Re: On-shoring of VMS support?? Message-ID: <1168425668.299754.164000@77g2000hsv.googlegroups.com> There was talk of changes here http://www.openvms.org/stories.php?story=06/12/01/4954974 ------------------------------ Date: 10 Jan 2007 08:05:31 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: On-shoring of VMS support?? Message-ID: In article <45A44C5E.384DE477@spam.comcast.net>, David J Dachtera writes: > > Now - how do we get front-page exposure to OpenVMS's woes? Maybe not a similar > response, but hey - no harm in trying, to my mind. Recipe for VMS exposure: 1) get rich 2) but Gartner 3) push VMS ------------------------------ Date: Wed, 10 Jan 2007 04:46:51 -0500 From: JF Mezei Subject: Star Trek and $CONFIGURE GALAXY Message-ID: <45a4b184$0$30646$c3e8da3@news.astraweb.com> Someone at VMS was clearly a Star Trek TNG fan. Stumbled on $CONFIGURE GALAXY , which brings up a GUI galaxy configuration utility. On the right is a big red button labeled "ENGAGE".... BTW, on a single CPU Alpha (DS10L), can one configure a GALAXY consisting of a single instance ? Does this utility bring anyinteresting GUI management to VMS on single CPU systems ? ------------------------------ Date: Wed, 10 Jan 2007 07:53:27 -0600 From: John Subject: Re: Star Trek and $CONFIGURE GALAXY Message-ID: <45A4EFD7.7010204@comcast.net> GALAXY is a SYSGEN parameter - and yes you can configure it as such for any version of VMS - though there is a license required for GALAXY - not sure how a lack of this license would affect a DS10L. However (and I assume this from experience with the GALAXY GUI) that the DS10L would only have a primary CPU and there would be nothing that could be done with it. JF Mezei wrote: > Someone at VMS was clearly a Star Trek TNG fan. > > Stumbled on $CONFIGURE GALAXY , which brings up a GUI galaxy > configuration utility. On the right is a big red button labeled > "ENGAGE".... > > BTW, on a single CPU Alpha (DS10L), can one configure a GALAXY > consisting of a single instance ? Does this utility bring > anyinteresting GUI management to VMS on single CPU systems ? > ------------------------------ Date: 10 Jan 2007 07:36:01 -0800 From: "Galen" Subject: Re: Star Trek and $CONFIGURE GALAXY Message-ID: <1168443361.835419.225210@i39g2000hsf.googlegroups.com> John wrote: > GALAXY is a SYSGEN parameter - and yes you can configure it as such for > any version of VMS - though there is a license required for GALAXY - not > sure how a lack of this license would affect a DS10L. > Even without a license, you can run a single instance Galaxy with just one CPU. > However (and I assume this from experience with the GALAXY GUI) that the > DS10L would only have a primary CPU and there would be nothing that > could be done with it. > Having GALAXY enabled on a single instance obviously doesn't buy you anything in terms of processing power, but it does enable at least some of the Galaxy related commands and programming features. This allows one, in at least a limited fashion, to design, test, and execute code that uses these features. > JF Mezei wrote: > > Someone at VMS was clearly a Star Trek TNG fan. > > > Stumbled on $CONFIGURE GALAXY , which brings up a GUI galaxy > > configuration utility. On the right is a big red button labeled > > "ENGAGE".... > > > BTW, on a single CPU Alpha (DS10L), can one configure a GALAXY > > consisting of a single instance ? Does this utility bring > > anyinteresting GUI management to VMS on single CPU systems ? ------------------------------ Date: 10 Jan 2007 12:37:00 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: Star Trek and $CONFIGURE GALAXY Message-ID: In article <45a4b184$0$30646$c3e8da3@news.astraweb.com>, JF Mezei writes: > Someone at VMS was clearly a Star Trek TNG fan. > > Stumbled on $CONFIGURE GALAXY , which brings up a GUI galaxy configuration > utility. On the right is a big red button labeled "ENGAGE".... > > BTW, on a single CPU Alpha (DS10L), can one configure a GALAXY consisting > of a single instance ? Does this utility bring anyinteresting GUI > management to VMS on single CPU systems ? Although I suspect there are some Star Trek fans in VMS Engineering, to me it just seems like another case of a easy to use VMS human interface. How many people on this planet will believe themselves to be technical enough to configure a VMS Galaxy and not understand the terminology on the GUI? Yes, I'm fairly sure you can configure a single instance Galaxy on your DS10L. I have plans to do so with mine just so I can get some Galaxy experience under my belt. The VMS 8.3 installation asked me if I wanted to do it, I just said no for the time being. I suspect it wouldn't have bothered asking if the system coudln't do it. (IIRC any Alpha or Itanium that can run VMS can do this). ------------------------------ Date: Wed, 10 Jan 2007 03:41:20 -0500 From: JF Mezei Subject: Strategy for common DECNET proxy files in cluster ? Message-ID: <45a4a22c$0$30586$c3e8da3@news.astraweb.com> What is the strategy for having a common DECNET proxy file in a cluster with multiple system disks ? What is the reason or having two files ? NET$PROXY.DAT;1 NETPROXY.DAT;1 I take it when moving those to a common area, both logical names need to be defined ? (netproxy and net$proxy) What is the role of each of those two files ? Or is one file sort of a relic ? (I have the real decnet (4) and none of my machines have ever had decnet 5 installed (the one that did was retired). Also, is there a way to have one entry for each user to work from any node to any node within the cluster ? Does this require setting a cluster alias? I have 7.3 on vax and 8.3 on alpha, would this be allowed without decnet routing licence ? ------------------------------ Date: Wed, 10 Jan 2007 07:32:30 -0700 From: "Michael D. Ober" Subject: Re: Stuck TCPIP connections Message-ID: <45a4f8fe$0$10302$815e3792@news.qwest.net> The main cause of this is a overloaded route to your news server. What happens is that the NNTP client loses the connection and then reconnects before the server times out the connection. Outlook Express is especially prone to this issue. Mike Ober. "JF Mezei" wrote in message news:ada2$45a425cb$cef8887a$28514@TEKSAVVY.COM... > Had problems with a news server yesterday. They must have changed > something at their end because I did not have it before. > > Basically, connections don't get properly shutdown and eventually the news > server complains I have more than 3 concurrent sessions going. > > I could then go to TCPIP DELETE DEVICE BGxxxx to zap the stray connections > and service would be restored for a few minutes untiol t happened again. > > Dowloading a binary item with a few images in it would result in something > like: > >> tcp 102 0 bike.52844 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 102 0 bike.52845 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52846 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52847 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52848 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52849 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52850 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52851 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52852 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52853 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52854 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52855 >> ispnews.usenetserver.com.119 CLOSE_WAIT >> tcp 56 0 bike.52856 >> ispnews.usenetserver.com.119 CLOSE_WAIT > > The thing is that the above connections were created more than 8 hours > ago. > > They also show up in TCPIP SHOW DEV > > From the usenetserver's point of view, those do not exist anymore because > it let me in now. > > > Question: shouldn't TCPIP connections eventually time out on their own and > go away ? > > I can understand some protocol or application errors resulting in a link > not being closed properly and remaining in CLOSE WAIT state. But shouldn't > that eventually time out and the connection be fully zapped ? ------------------------------ Date: 10 Jan 2007 06:44:56 -0800 From: mckinneyj@saic.com Subject: Re: Stuck TCPIP connections Message-ID: <1168440296.125689.252390@i56g2000hsf.googlegroups.com> JF Mezei wrote: > Had problems with a news server yesterday. They must have changed something > at their end because I did not have it before. > > Basically, connections don't get properly shutdown and eventually the news > server complains I have more than 3 concurrent sessions going. > > I could then go to TCPIP DELETE DEVICE BGxxxx to zap the stray connections > and service would be restored for a few minutes untiol t happened again. > > Dowloading a binary item with a few images in it would result in something > like: > > > tcp 102 0 bike.52844 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 102 0 bike.52845 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52846 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52847 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52848 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52849 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52850 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52851 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52852 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52853 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52854 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52855 ispnews.usenetserver.com.119 CLOSE_WAIT > > tcp 56 0 bike.52856 ispnews.usenetserver.com.119 CLOSE_WAIT > > The thing is that the above connections were created more than 8 hours ago. > > They also show up in TCPIP SHOW DEV > > From the usenetserver's point of view, those do not exist anymore because > it let me in now. > > > Question: shouldn't TCPIP connections eventually time out on their own and > go away ? > > I can understand some protocol or application errors resulting in a link > not being closed properly and remaining in CLOSE WAIT state. But shouldn't > that eventually time out and the connection be fully zapped ? There is no timeout for CLOSE_WAIT state sockets - they will remain until your local application either closes the socket or the image holding them open runs down. Seeing a socket in this state indicates that your TCP stack has received a close request from the remote end but your local software hasn't reciprocated. In this instance since it appears that the need to close the sockets will never be satisfied, therefore it is likely that you have a bug in your local software. ------------------------------ Date: 10 Jan 2007 08:43:53 -0800 From: "mendondude@gmail.com" Subject: Re: VMS732_UPDATE-V0900 fails Message-ID: <1168447433.154255.291620@o58g2000hsb.googlegroups.com> On Jan 9, 9:32 pm, David J Dachtera wrote: > > About the only ones I didn't blow through the roof were WSextent and Pgflquota. > > Anyone have any ideas? (V7.3-2, no prior ECOs except PCSI V0300, the machine has > 256MB RAM) > Pagefile quota. Jack that to 200000 or so. 50000 is much too small ------------------------------ End of INFO-VAX 2007.020 ************************