[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ssdevo::hsj40_product

Title:HSJ30/40 Product Conference
Moderator:SSDEVO::EDMONDS
Created:Mon Jul 12 1993
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:1264
Total number of notes:4958

1229.0. "Stupid Pet HSJ tricks." by NABETH::alan (Dr. File System's Home for Wayward Inodes.) Tue Apr 08 1997 23:52

	Would those appropriately knowledgable please comment on the
	wisdom (or lack of it) of the following:

	1.  Initialize a VMS disk on a local SCSI controller (or in my
	    particular version of the experiment a TRANSPORTABLE disk
	    on an HSJ).  Mount and copy some files to it.  Dismount and
	    set aside.

	2.  Add a disk to a shelf on an HSJ as NOTRANSPORTABLE.  Add
	    a unit:

		ADD disk disk140 1 4 0 NOTRANSPORTABLE
		ADD d140 disk140

	    Remove the disk and set it aside.

	3.  Put the TRANSPORTABLE (or local controller INITed) disk
	    in the slot of the NOTRANSPORTABLE disk.

	4.  Mount the VMS file system and use "normally".

	I'll point out that I did this earlier to night and for the
	short time that I was using it, nothing bad happened.

	At some point in the future I would expect one of two things
	to go disasterously wrong, but if this is safer than it looks
	please say so:

	Diaster #1: VMS thinking the disk is larger than it actually
	is (due to the metadata space not being present in the TRANSPORTABLE
	version), tries to use that space, only to be told by the device
	that the space doesn't exist.

	Disaster #2: The HSJ needing to use the metadata space of what
	it believes is a NOTRANSPORTABLE disk, gets bent way out of
	shape when it find the metadata space isn't.
T.RTitleUserPersonal
Name
DateLines
1229.1Lack of it (wisdom)SSDEVO::JACKSONJim Jackson, HSJ40 RAID teamWed Apr 09 1997 10:5819
If you managed to get a TRANSPORTABLE disk to work in a NOTRANSPORTABLE
slot, it was only because at one time that disk had been initialized as
NOTRANSPORTABLE.  If the metadata is not on the media, we don't allow a
NOTRANSPORTABLE disk to go online.

** You're playing with fire here. **

In its life as a TRANSPORTABLE disk, some of the old metadata could have
been altered.  This could cause unpredictable results when you use it as a
TRANSPORTABLE disk.

The VMS filesystem attempts to use blocks near the directory (near the
center) first, so the blocks at the end that "no longer exist" are among the
last blocks used.  Thus, your observation that things appear to work.
However, because the controller is now using some disk space for its own
purposes, if this disk ever goes back to a TRANSPORTABLE slot, some of the
user space has been altered.

Your expectations of disaster are not out of line.
1229.2Its the not-VMS cases that worry me!, anyone tried an HSx with the E2FS ?SSDEVO::FIALAMe, I'm just a recycler.Wed May 28 1997 10:2221
When you VMS init a disk it fixes up the SCB [sorta superblock for the Unix impaired]
and at that time the disk size is "frozen".
If you later expand the size by nefarious means VMS wont care in the least
as long as you keep using the filesystem. If you circumvent the filesystem
the "extra" space can be gotten at.
[To prove this idea of a reduced device try a backup/restore of a small
DUAx device to a larger variant]...

Initializing a large device then reducing its apparent size is caught by
MSCP with an invalid LBN when VMS tries to access the data off the end.
So you could! create a device with a filesystem that has inaccessable
files this way [why, oh why?].

The problem [or fire danger] is that when the unit is converted from
with-metadata to without, the metadata is not destroyed. Consequently you
could flame-it then resurect it as a phoenix in another life.

If you are really adventurous you could move devices between disparate
controller architectures also. :-)

BTW, none of the above is advised, supported... its sure to lead to disaster.