| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 3762.1 | I'm not sure that the explanation fits... | STEVMS::PETTENGILL | mulp | Fri Nov 22 1996 22:00 | 6 |
| 3762.2 | Further clarification of VMS bug | COMICS::WEIR | John Weir, UK Country Support | Wed Dec 04 1996 05:07 | 109 |
| 3762.3 | Why no automatic reboot? | UTRTSC::VROOIJ | Learning to fly... | Fri Dec 06 1996 07:43 | 12 |
| 3762.4 | patch kits? | DBOT09::Barrell | Steve Barrell, Dublin. DTN: 827-2267 | Mon Jan 27 1997 11:58 | 10 |
| Any pointers to patch kit(s)?
I searched through TIMA looking for "ctlpages", "system_primitives", "UVO104126"
and "CFS.38366", but could not find anything that seemed to match this
CTLPAGES - CTLIMGLIM problem.
Some internal systems in the US are migrating from Phase IV to DECnet/OSI
and they have CTLPAGES = 700 for some reason.
|
| 3762.5 | VAXSYS08_062, I think | COMICS::WEIR | John Weir, UK Country Support | Thu Jan 30 1997 04:16 | 10 |
|
Steve,
The workaround is to increase CTLIMGLIM to something just below CTLPAGES
(eg to 60 pages less).
The fix will appear one day... For VMS V6.2, I think it will be VAXSYS08_062
John
|
| 3762.6 | | TITAN::Barrell | Steve Barrell, Dublin. DTN: 827-2267 | Thu Jan 30 1997 07:08 | 10 |
| Thanks John,
Folks are doing the workaround, but with many [new] contractors around with
various skill levels, I was hoping to make life just a little easier by putting
in a patch - if and when it's available.
I'll keep an eye open for VAXSYS08 or something of that ilk.
Regards,
Steve.
|
| 3762.7 | Could this happen at certain time intervals? | TFOSS1::HEISER | Maranatha! | Thu Mar 13 1997 11:06 | 21 |
| Is there a specific timeframe surrounding when this VA expansion could
occur ala the Pathworks or VMS delta time blitzes? Yesterday morning, I
had 9 different nodes in 5 different clusters experience:
- NET$ACP consuming greater than 50% of the CPUs' time when I finally
logged in. Consumption was on a gradual upswing.
- Network Partner Exited for any attempted DECnet connection to the
affected system.
- When I grabbed the console and finally got in, I could make outgoing
DECnet connections, but incoming connections still get Network Partner
Exited.
All nodes had CTLPAGES at several hundred > CTLIMGLIM. All nodes
displayed the exact same symptoms within 3 hours of each other. These
nodes were in different clusters and geographically separated. The
timing makes me really suspicious. Almost looked like a virus ;-)
thanks,
Mike
|
| 3762.8 | Not time related; check DNS Servers | COMICS::WEIR | John Weir, UK Country Support | Fri Mar 14 1997 03:21 | 20 |
|
Mike,
I have never noticed anything to make several nodes affected at the same time,
and there is NO reason to believe that there is anything time related (it's
a very simple coding error, and nothing to do with time).
But... The bit of DECnet/OSI code affected by this VMS bug is the DNS Clerk,
so, if you had a DNS namespace problem (eg some servers went into an unusual
state) then the DNS Clerk might have to work harder and allocate more memory,
and this might be enough to expose the problem. It should be noted that the
onset of the problem has always been unpredictable, as it has depended upon the
usage and fragmentation of the CTL pool. Thus, as implied above, different
levels of Clerk activity will affect pool usage and fragmentation and may
make the problem appear or disappear...
Regards,
John
|