|
HP OpenVMS Systemsask the wizard |
|
The Question is: our company just upgraded to version 7.3 we have 3 nodes on 1 cluster. on 1 node we do a type "filename.txt" and it displays data correctly. on another node we do a type "filename.txt" and it shows some other data. filename.txt is the same file! what is the fix? The Answer is :
The OpenVMS Wizard would expect that the files are not actually
the same file, or that this is potentially related to known
problems within the OpenVMS Alpha V7.3 XFC cache.
You will want to disable use of XFC on OpenVMS Alpha V7.3 per
the available documentation, pending release of V2.0 of the XFC
ECO kit for V7.3, or later -- topic (7408) is related.
Please use the $ DIRECTORY /FILE_ID command on each host or each
process that is referencing the file. This command will display
the file identification (FID), a value which is a unique reference
to the contents of a file on a particular disk or on a particular
shadowset.
Some of the more common causes of the display of different data
for the same filename involve differences in the file specification
resulting from logical name translations, and differences in the
process default device specification. These translation differences
can involve search list logical names, and -- depending on the
particular filename specification syntax used -- to logical names
which match the name of the particular file.
If the file identification value differs (and the physical device
is the same device), then the file differs.
One of the more common situations involves the system disk in an
OpenVMS Cluster configuration, and specifically copies of files
located in the SYS$SPECIFIC: and SYS$COMMON: roots. If there is
a copy of the file in SYS$SPECIFIC: on one host and another copy
of the file in SYS$COMMON:, the behaviour you describe is expected.
You can use the SHOW LOGICAL command to examine the various components
of the file specification; to display any logical name translations.
Be certain to carefully evaluate potentially subtle differences, such
as differences in the device name of the disk, or differences in the
name of the system disk system root.
In the following example, any user SET DEFAULT to SYS$SYSTEM would
see the contents of two directories -- SYS$SYSTEM is a search list
specifying the "HOST$DKA0:[SYS0.]" (SYS$SPECIFIC) system disk root
and the "HOST$DKA0:[SYS0.SYSCOMMON.]" (SYS$COMMON:) root.
$ show logical sys$system
"SYS$SYSTEM" = "SYS$SYSROOT:[SYSEXE]" (LNM$SYSTEM_TABLE)
$ show logical sys$sysroot
"SYS$SYSROOT" = "HOST$DKA0:[SYS0.]" (LNM$SYSTEM_TABLE)
= "SYS$COMMON:"
1 "SYS$COMMON" = "HOST$DKA0:[SYS0.SYSCOMMON.]" (LNM$SYSTEM_TABLE)
Other potential causes can include use of allocation class zero
on two cluster member nodes -- both references (only) appear to
be to the same storage device. (A correct cluster configuration
can have overlaps in the disk device names in allocation class
zero, though all disk storage devices within a non-zero allocation
class should be unique across all cluster members.)
The OpenVMS Wizard strongly recommends review of the OpenVMS User's
Guide, and particularly the documentation on the format and use of
filenames, and the documentation on logical names.
|