tencati@JPL-VLSI.ARPA.UUCP (07/31/86)
I recently upgraded to VMS V4.4 on my 11/780, and I noticed a "new" behavior... If a user does a SET HOST 0 to make a connection back to this machine, then logs in and runs EDT, the device RTA1: begins logging errors. These show up on any terminal from which a SHOW ERROR is typed. The more the user types into EDT, the more errors that occur. When the user logs off the decnet process, the device errors on RTA1 are erased, and that device no longer shows up in the SHOW ERROR display. Joe Leonard reported his V4.4 crashing when he was in EDT, this may be a related problem, or it could be a strange new quirk. Any insight would be appreciated. Ron Tencati System Mgr/JPL-VLSI.ARPA Jet Propulsion Laboratory Pasadena, Ca.
McGuire_Ed@GRINNELL.MAILNET (08/04/86)
>Date: Thu, 31 Jul 86 09:09:42 PDT >From: tencati@Jpl-VLSI.ARPA >Subject: VMS 4.4: Device Errors > >If a user does a SET HOST 0 to make a connection back to this machine, then >logs in and runs EDT, the device RTA1: begins logging errors. Same thing has been happening here under V4.4 on an 8600. I have no idea why.
LEICHTER-JERRY@YALE.ARPA (08/13/86)
I recently upgraded to VMS V4.4 on my 11/780, and I noticed a "new"
behavior...
If a user does a SET HOST 0 to make a connection back to this machine,
then logs in and runs EDT, the device RTA1: begins logging errors. These
show up on any terminal from which a SHOW ERROR is typed....
Joe Leonard reported his V4.4 crashing when he was in EDT, this may be a
related problem, or it could be a strange new quirk. Any insight would be
appreciated.
Here's the scoop, as explained to me by a contact at DEC: RT's are software-
only devices, so the error count in the UCB does not indicate hardware prob-
lems (like timeouts), as is the case for devices that connect to real hard-
ware. Instead, the error count is used by the RT driver to report protocol
errors. The errors you are seeing are the result of a bug introduced in
VMS V4.4. They have NO functional significance, and do not affect communi-
cation - or system operation - in any way.
Note that RT's are cloned UCB's - the only permanent UCB is RTA0, which is
copied to create RTAn when a new RT is needed. When RTAn is no longer needed,
the cloned UCB is destroyed, and the memory it used returns to non-paged pool.
Any RTAn's created later are also created from RTA0; even if they get the same
"n" as a previously-used RTAn, they have no connection to it. That's why the
error count disappears when you finish with the link.
The bug that causes the "errorneous error" reports will be fixed in a future
release of VMS after VMS V4.5. Until then, just ignore the errors.
-- Jerry
-------