1 INFO-VAX	Sat, 24 Dec 2005	Volume 2005 : Issue 715       Contents:# Addendum: TCPIP5.4  patch 5 problem ' Re: Addendum: TCPIP5.4  patch 5 problem  DUMP/OUTPUT strangeness  Honeypot stats/ Re: HP : Massive strike and protest march today  MERRY  VAXMAS !  :-)) Re: Question about large numbers of Locks ) Re: Question about large numbers of Locks ) Re: Question about large numbers of Locks ) Re: Question about large numbers of Locks 4 Re: Some people are willing to support their product Re: SSH vs Telnet  TCPIP5.4  patch 5 problem   F ----------------------------------------------------------------------  % Date: Sat, 24 Dec 2005 08:51:44 -0800 # From: "Tom Linden" <tom@kednos.com> , Subject: Addendum: TCPIP5.4  patch 5 problem( Message-ID: <ops2arsiq7zgicya@hyrrokkin>  F On Sat, 24 Dec 2005 08:36:59 -0800, Tom Linden <tom@kednos.com> wrote:  = > I have 5.4 running on 7.3, 7.3-1 and 7.3-2  and I installed 9 > said patch to each of them, but 7.3 doesn't like it :-(  > F > on 7.3 I run the POP server and  MX V5.4  After installing the patchD > I can no longer connect using SSH and when I try to send mail fromI > Outlook client I get the error message Unable to connect to the server.  > 0 > When I rebooted the following message appeared > K > -TCPIP-E-FACNOTSTART, attempt made to run a TCPIP image before starting    > the facility6 > %TCPIP-I-ANA_COMPLE, ANALYZE completed on host FREJA > H > Not sure what caused that the only diff in SYSTARTUP_VMS.COM was the   > addition of the 9 > second line below in accordance with the release notes.  > $ > $       @sys$startup:tcpip$startup5 > $       INSTALL ADD SYS$SHARE:DECW$SETSHODISSHR.EXE  > K > If this is an imcompatibility with MX then how do I back the patch back    > out?  I I sopped MX manually and restarted and mail now works, so what is wrong    with+ the startup sequence which reads as follows  $!      Start tcpip services? $       DEFINE/SYSTEM     TCPIP$POP_USE_MAIL_FOLDER        TRUE ? $       DEFINE/SYSTEM     TCPIP$POP_PERSONAL_NAME          TRUE @ $       DEFINE/SYSTEM     TCPIP$POP_POSTMASTER             "TOM"? $       DEFINE/SYSTEM     TCPIP$POP_IGNORE_MAIL11_HEADERS  TRUE < $       DEFINE/SYSTEM     TCPIP$SMTP_JACKET_LOCAL          1" $       @sys$startup:tcpip$startup3 $       INSTALL ADD SYS$SHARE:DECW$SETSHODISSHR.EXE : $       IF F$SEARCH("SYS$STARTUP:MX_STARTUP.COM") .NES. "" $       THEN4 $               @SYS$STARTUP:tcpip$smtp_shutdown.com@ $               DEFINE/SYSTEM/EXEC MX_INVALID_DOMAIN_ADDRESSES   "64.94.110.11"@ $               DEFINE/SYSTEM/EXEC MAIL$PROTOCOL_SMTP MX_MAILSHR; $               DEFINE/SYSTEM/EXEC MX_PROTOCOL_PREFIX SMTP% < $               DEFINE/SYSTEM/EXEC MX_FROM_HOST "kednos.com"+ $               @SYS$STARTUP:MX_STARTUP.COM 
 $       ENDIF < $       @SYS$SYSDEVICE:[VMS$COMMON.SYSMGR]TCPIP$FTP_SHUTDOWN, $       @disk$common:[hg.COM]FTP_STARTUP.COM   ------------------------------  % Date: Sat, 24 Dec 2005 12:35:39 -0500 % From: BRAD <bradhamilton@comcast.net> 0 Subject: Re: Addendum: TCPIP5.4  patch 5 problem* Message-ID: <43AD86EB.9080704@comcast.net>   Tom Linden wrote: H > On Sat, 24 Dec 2005 08:36:59 -0800, Tom Linden <tom@kednos.com> wrote:K > I sopped MX manually and restarted and mail now works, so what is wrong    > with- > the startup sequence which reads as follows  > $!      Start tcpip servicesA > $       DEFINE/SYSTEM     TCPIP$POP_USE_MAIL_FOLDER        TRUE A > $       DEFINE/SYSTEM     TCPIP$POP_PERSONAL_NAME          TRUE B > $       DEFINE/SYSTEM     TCPIP$POP_POSTMASTER             "TOM"A > $       DEFINE/SYSTEM     TCPIP$POP_IGNORE_MAIL11_HEADERS  TRUE > > $       DEFINE/SYSTEM     TCPIP$SMTP_JACKET_LOCAL          1$ > $       @sys$startup:tcpip$startup  H It's been a few years since I ran TCP/IP Services for OpenVMS, but ISTM H that the DEFINE/SYSTEM commands above should happen _after_ the startup H of TCP/IP Services.  I seem to recall that there is a <mumble>SYSTARTUP H command file for TCP/IP Services, where these commands could reside.  I G _think_ that TCPIP$STARTUP calls this file near the end of its startup  	 procedure    <snip>   ------------------------------  + Date: Sat, 24 Dec 2005 18:24:58 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)  Subject: DUMP/OUTPUT strangeness$ Message-ID: <dok3pp$m9v$1@online.de>   If I do       $ DUMP *.HTML  # then I get what I expect.  If I do        $ DUMP *.HTML/OUTPUT=X.X   ; then X.X contains only the dump of the last file.  If I do         $ DEFINE/USER SYS$OUTPUT X.X     $ DUMP *.HTML  < then I get the same strange result.  This works as expected:      $ PIPE *.HTML > X.X   What am I missing?   ------------------------------  % Date: Sat, 24 Dec 2005 12:24:15 -0500 ) From: "Neil Rieck" <n.rieck@sympatico.ca>  Subject: Honeypot stats 9 Message-ID: <Gufrf.5106$%N1.758532@news20.bellglobal.com>   K This post has nothing to do with OpenVMS but I thought I'd pass it on to my J security-minded peers after reading the "Christmas 2005 Wishes" from Bob @D InstantWhip. The following information was just pasted from an email0 currently circulating at my place of employment.  
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada.8 http://www3.sympatico.ca/n.rieck/links/cool_openvms.html   #####   F These Honeypot links refer to break-in stats from the newest availableH distributions of various operating systems. The software is installed onI systems which are quietly connected to the internet without an associated H DNS entry. The systems do nothing other than watch for, and record, hack0 attempts. The timer begins after "I/P discovery"  " During the November-2004 go round:,   Windows-XP SP1 was hacked within 4 minutes&   Red Hat Linux was hacked in 3 months  3 http://www.eweek.com/article2/0,1759,1752343,00.asp - http://www.techweb.com/wire/security/54201306 - http://www.techweb.com/wire/security/56200327   http://www.tracking-hackers.com/" http://www.honeynet.org/challenge/  L In September of 2005 I attended in IEEE tour of Raytheon Canada in Waterloo.F You should have seen the bewildered look on U of W (a big MS fan club)K students' faces when the Raytheon engineers told them that Windows security H was a joke and no company would seriously consider it for anything other" than a personal productivity tool.   ------------------------------    Date: 23 Dec 2005 22:55:08 -0800$ From: "AEF" <spamsink2001@yahoo.com>8 Subject: Re: HP : Massive strike and protest march todayC Message-ID: <1135407308.737961.124970@g43g2000cwa.googlegroups.com>   
 Lurker wrote: A > "David J Dachtera" <djesys.nospam@comcast.net> wrote in message ' > news:439B3628.36213F03@comcast.net...  > A > I know I've promised to quit this thread but I couldn't resist.  > < > So how do you feel about that David after the NY transport: > strike? And the fact that NY Times summed up the popular@ > feelings by the terse but explicit "You Rats!" And maybe, just; > maybe, you have happened to be in that city and trying to  > catch a bus?   Well, since you brought it up:  E That was the NY Post that had the full page headline "YOU RATS!". All G the major NY papers had editorials denouncing the TWU local -- even the 	 NY Times.   C I admit that they work under lousy conditions (but so do some other G workers) and they seem to have very strict work rules, but they did NOT B make that a big issue. The big issue was changing the pension fromD 25/55 to 30/62 or something like that, but only for NEW HIRES. So weB see that at least in this case a union cares for new hires, but ofA course that's not only the case. And they have great benefits and G pensions that most people can only dream of. The reality is that health D and pensions costs are rising rapidly, partly due to the cost of newC medical treatments and partly due to increased life expectancy. The F current benefit levels are not sustainable in the long run and the MTAE was trying to nip it in the bud. Of course this looks like a giveback C to the union and so they got upset. But there are 210 commuters per G transit worker. I really don't think it's fair to punish the commuters. C The union and its workers will likely have to pay some stiff fines: D $3000000 for the union and 6 days' pay for the workers, plus any law5 suits, plus any fines for being in contempt of court.   F Some of the transit workers are great workers. Some of them don't give a damn and are very unpleasant.   E And of course they don't give a damn about the massive commuter chaos A they caused. A fireman riding a bike to work was hit by a bus and ( nearly died. He's in critical condition.  : People missed their chemotherapy and radiation treatments.  A An interesting thing about this is the the TWU's parent union was F against the strike. They first told them not to strike. They they told them to go back to work.  F 7,000,000 commuters (that's 210 per transit worker) had to either walkF huge distances during a cold spell, pay high rates for taxis, or get 4G people in a car to be allowed into Manhattan south of 96th street. Many B businesses, during the height of the Christmas (holiday?) shoppingG season, endured heavy losses. Restaurants and museums suffered greatly. G Almost everyone, even most of the workers themselves, are glad they are 
 back to work.   G The NY Post dissected a TWU commercial and explained how it was full of  lies.   E I could go on and on, but I'd better stop here. I'm sure one can look  this stuff up on the Web.    AEF    ------------------------------  % Date: Sat, 24 Dec 2005 18:47:07 +1100 $ From: Phaeton <phaeton@iinet.net.au> Subject: MERRY  VAXMAS !  :-) J Message-ID: <43acfcf7$0$14676$5a62ac22@per-qv1-newsreader-01.iinet.net.au>  7 	I wish everyone a Merry VAXmas and an Alpha New Year !   E                                                       Cheers,   Csaba   E --------------------------------------------------------------------- F   CSABA I. HARANGOZO  |d|i|g|i|t|a|l|  phaeton at iinet dot net dot auE --------------------------------------------------------------------- <     EARTH::AUSTRALIA:[SYDNEY]HARANGOZO.CSABA;1, delete? [N]:   ------------------------------  + Date: Sat, 24 Dec 2005 13:30:04 +0000 (UTC) 7 From: moroney@world.std.spaamtrap.com (Michael Moroney) 2 Subject: Re: Question about large numbers of Locks( Message-ID: <dojigs$996$1@pcls4.std.com>  A Considering that the file access already uses lots of (RMS) locks H (as RMS coordinates access to the file), switching over to a lock-based H method won't increase lock usage and may actually _decrease_ lock usage.  K Clusterwide locks _may_ lose the most recent value of the lock value block  H if the wrong node crashes at the wrong time, if that's important to you.E You will lose it if the whole cluster is down at once, while the file 1 will survive.  Again, if that's important to you.   G How do you check for updates, periodic polling?  How often? If frequent J polling is what you're doing, locks with blocking ASTs will be a big win, . if you're running out of CPU or I/O bandwidth.  I If you are running out of CPU and are trying to avoid a hardware upgrade, G it's likely worth fixing, but if not, it may be best not to tamper with 3 something that works that your 'upgrade' may break.    ------------------------------    Date: 24 Dec 2005 08:08:10 -0800 From: rcbryan@hotmail.com 2 Subject: Re: Question about large numbers of LocksC Message-ID: <1135440490.517384.206320@f14g2000cwb.googlegroups.com>   C I had a version where I mapped the file into a shared section using F mmap but updates were not moved to other nodes on the cluster unless IC closed the file.  As I think about it now, it occurs to me that the F changes of state are all done from one user program (possibly multipleC copies running on multiple nodes but not working on the same record E simultaneously).  It would be reasonable to close the file and reopen C it each time I make an update and expect the updated values to make D their way to disk and then to other nodes.  (I will have to try that when I get back to work.)   G I guess one question is:  Is the VMS lock mechanism is bright enough to F where there is not fixed CPU overhead for each lock?  If they just sitE there until there are changes than the CPU overhead would be minimal.   E One of the processes regularly uses over an hour of CPU time per day, G sometimes closer to two hours.  My estimate (using PCA on a development C system) is that about 20% of that is reading the file.  Most of the E processes do not have anywhere near this amount of activity but a few  do.   = "If it ain't broke..." I have little patience with people who A noodle things up for the sake of "better programming" (or vanity) F but in this case, one of the managers asked if I can see where some ofF the CPU time is going and do something about it.  This is a case whereE a change can be made to in a small place and have a big payoff. There C are a lot of other egregious abusers of CPU time but they are a lot B harder to address.  They paid the big bucks a while ago to get theF GS1280s and they want to keep on eye on the CPU time.  I think that ifE they eliminated all the wasted CPU time on the system, they could run A on an ES40 but it is cheaper and less risky to just buy hardware.    Thanks for your input.	 /RC Bryan    ------------------------------  % Date: Sat, 24 Dec 2005 11:46:43 -0500 ) From: "Neil Rieck" <n.rieck@sympatico.ca> 2 Subject: Re: Question about large numbers of Locks9 Message-ID: <uXerf.5088$%N1.756609@news20.bellglobal.com>   ' <rcbryan@hotmail.com> wrote in message  = news:1135440490.517384.206320@f14g2000cwb.googlegroups.com... D >I had a version where I mapped the file into a shared section usingH > mmap but updates were not moved to other nodes on the cluster unless IE > closed the file.  As I think about it now, it occurs to me that the H > changes of state are all done from one user program (possibly multipleE > copies running on multiple nodes but not working on the same record G > simultaneously).  It would be reasonable to close the file and reopen E > it each time I make an update and expect the updated values to make F > their way to disk and then to other nodes.  (I will have to try that > when I get back to work.)  >  [...snip...] >  > Thanks for your input. > /RC Bryan  > M As others have already pointed out, spin-locking on a file is not a big deal  L on a lightly loaded system especially if that file is used for some kind of I limited inter-process communication. However, when your system gets busy  G you'll will want to squeeze every last drop of performance out of your  I programs and one better way to do locks on a cluster (where caching file  K caching has limited effects) is to interact directly with the built-in DLM  K (Distributed Lock Manager). Check out the following link for a little more   information and a demo. 4 http://www3.sympatico.ca/n.rieck/docs/DLM_notes.htmlJ I'm not saying this method will be easy but I have found the effort worth  while.  
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada.8 http://www3.sympatico.ca/n.rieck/links/cool_openvms.html   ------------------------------  % Date: Sat, 24 Dec 2005 18:33:40 -0000 @ From: "Alex Daniels" <alexNOSPAMHERETHANKSdaniels@themail.co.uk>2 Subject: Re: Question about large numbers of Locks4 Message-ID: <43ad9487$0$912$db0fefd9@news.zen.co.uk>  M "Colin Butcher" <colin_DOT.butcher_AT@xdelta_DOT.co_DOT.uk> wrote in message  7 news:RBfrf.14352$iz3.2117@text.news.blueyonder.co.uk... L > Are you using Galaxy / shared memory with instances running in partitions?F > If so then consider a 'galactic memory' global section - which will 	 > survive H > a reboot provided that at least one instance using the global section  > stays  > up.   E A neat solution, but if he wants to go to Itanium at some point that  ' implementation might present a problem.   I I've been told by HP shared memory is not going to be implemented in the  J initial release of vpars (which won't happen until post Montecito), and I K have not seen any commitment for shared memory to ever be (and I did ask),  $ although it wasn't ruled out either.   Alex     ------------------------------  % Date: Sun, 25 Dec 2005 00:46:27 +0800  From: prep@prep.synonet.com = Subject: Re: Some people are willing to support their product - Message-ID: <87y82abfz0.fsf@prep.synonet.com>   ) "Main, Kerry" <Kerry.Main@hp.com> writes:   , > Here's a thought to spice up the thread ..   > :-)   C > Another look at the same issue of cost is "how many Customers can " > really afford Windows or Linux?"  E > If one looks at the effort, resources and $'s required to QA / test > > monthly security patches against your main applications with7 > Windows/Linux, can you really afford these platforms?   B > Course, you could decide to roll-out monthly OS security patchesF > without any testing, but then you are not likely a shop with serious& > uptime or availability requirements.  E > If one considers that the average x86 server in production today is F > 10-20% utilized at peak times, how long can your company continue to0 > support the one-application, one server model?  E Have you quit your job? No one fom HP is allowed to even THINK things  like that...   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------    Date: 24 Dec 2005 10:24:15 -0800, From: "rcyoung" <rcyoung@aliconsultants.com> Subject: Re: SSH vs TelnetC Message-ID: <1135448654.923243.168210@g49g2000cwa.googlegroups.com>   F I'm not sure if you can adjust PuTTY for what you want... but there isE another option...use PuTTY/ssh to redirect ports so that you set up a F "secure tunnel" between the vms and your w2k for the telnet port. Then continue to use telnet.    Look at   V http://www.ssh.com/support/documentation/online/ssh/adminguide/32/Port_Forwarding.html   ------------------------------  % Date: Sat, 24 Dec 2005 08:36:59 -0800 # From: "Tom Linden" <tom@kednos.com> " Subject: TCPIP5.4  patch 5 problem( Message-ID: <ops2aq3xh3zgicya@hyrrokkin>  ; I have 5.4 running on 7.3, 7.3-1 and 7.3-2  and I installed 7 said patch to each of them, but 7.3 doesn't like it :-(   D on 7.3 I run the POP server and  MX V5.4  After installing the patchB I can no longer connect using SSH and when I try to send mail fromG Outlook client I get the error message Unable to connect to the server.   . When I rebooted the following message appeared  I -TCPIP-E-FACNOTSTART, attempt made to run a TCPIP image before starting    the facility4 %TCPIP-I-ANA_COMPLE, ANALYZE completed on host FREJA  F Not sure what caused that the only diff in SYSTARTUP_VMS.COM was the   addition of the 7 second line below in accordance with the release notes.   " $       @sys$startup:tcpip$startup3 $       INSTALL ADD SYS$SHARE:DECW$SETSHODISSHR.EXE   I If this is an imcompatibility with MX then how do I back the patch back    out?   ------------------------------   End of INFO-VAX 2005.715 ************************