Product: AMASS for UNIX® Version 5.4
Operating Systems: IBM® AIX 5.1 and 5.2 (32-bit only)
HP Tru64™ UNIX 5.1, 5.1A, and 5.1B
HP-UX® 11.00 (32-bit and 64-bit)
HP-UX® 11i (11.11) (32-bit and 64-bit) (PA-RISC only)
SGI IRIX® 6.5.16m - 6.5.23
Sun Solaris™ 8 and 9 (32-bit and 64-bit) (SPARC™ only)
Date: December 2004
Contents
PageTopic
1Purpose of this Release
2New Features
5Additional Support
6Software Installation
6System Requirements
8Compatibility Matrix
10Tape Features
12Upgrade Guidelines
12Upgrade Instructions
13Authorization String
14Fibre Channel Guidelines
16Operating Guidelines
17Known Issues
19
20
Resolved Issues
Documentation
Purpose of this Release
This release adds several new features and enhancements that extend the capabilities of AMASS for UNIX.
These release notes list corrections to problems that were encountered in previous releases of AMASS for
UNIX. Visit www.adic.com
for additional information about AMASS for UNIX and previous releases.
These features and enhancements are new in this release of AMASS for UNIX:
•AMASS Read Only—AMASS read only allows users to set AMASS to a read only state. You can
read from AMASS, but not write or delete. To change the state to read only we use the amasstat
command. The options are as follows:
Usage: amasstat [-aicfswuy] [-t sec]
•-a Changes filesystem to active status.
•-i Changes filesystem to inactive status (reads and writes from/to open files continue).
•-r Changes filesystem status to read only.
•-R Changes filesystem status to read/write.
•-c Checks current status, but do not change.
•-f Forces status change even if timeout occurs.
•-s Suppresses messages from kernel.
•-t Time, in seconds, to wait for AMASS to go inactive.
•-u Displays this usage message.
•-v Displays the current AMASS version.
•-w Disables writes to AMASS when inactive (reads from open files continue).
•-y Suppresses confirmation and informational messages.
If -a, -i, and -c are not specified, then the active status of the amass filesystem is toggled.
•Cache Size—AMASS supports up to 256 partitions and has been certified with up to 4 TB cache.
The 4 TB cache budget may be spent on a single partition or dispersed among multiple partitions.
In theory, each partition could use 4 TB cache, or up to 64 PB total; however, the total amount of
cache among all partitions has only been certified up to 4 TB.
•drivestat—Three new command line options have been added to this tool.
•-r Resets device runtime state flags to initial condition.
•-j Resets device eject state flags to initial condition.
The -rj option addresses a drive in an active state, but has no work scheduled for the drive.
This situation arises after a volume, stuck in a drive, has been cleared offline. After restoring
the drive to an active state, the drive remains idle. The -rj option will remove the drive state
at the time of failure and restore the drive to its initial conditions. AMASS will assume and trust
the operator to have cleared the drive and put the media away.
The -rj option cannot be run on an active drive.
•-D Displays detailed amass debug logging.
The -D option has been a hidden option for several release cycles and one AMASS support
person has instructed users how to use it. It is now part of the standard -u usage output.
Usage: drivestat -D <device-number> <juke-number> <debug-option>; where <debug-option> is defined as the arithmetic sum of the following:
•8 = Sets select_debug flag (in libsched select_drive function).
•15 = ALL: 1+2+4+8
•Enhanced Cache—There are enhancements to the AMASS feature Enhanced Cache Control. The
core of AMASS is a cache in which data is held prior to migrating to/from media. The cache is broke
up into cache blocks whose size is configurable. Without the Enhanced Cache Control feature
turned on, AMASS reuses cache blocks which are the oldest on the cache.
With Enhanced Cache Control the cache blocks are allocated to cache block lists. Each of these
lists can then be weighted to control from where cache blocks are acquired. The cache block lists
are then associated with AMASS volume groups. AMASS volume groups allow users to group data
in a AMASS directory with a set of volumes in the tape/optical library. When associating a volume
group to a cache block list, the user can do so for read or writes. So once a file is written to cache,
it is then migrated to tape, and could be made to persist in cache by having the file's volume group
be associated with a highly weighted cache block list.
•Eight cache block priority levels are maintained within the AMASS files system.
•Cache block reallocation priorities are established on a volume group basis. No additional
controls are provided for individual directories or files.
•For each volume group, two cache list priorities (one through eight) are assigned. One list is
assigned for reads and the other list is assigned for writes.
•For a particular volume group, the priority can be changed dynamically for the read/write
priority, but the new priority takes effect for cache blocks accessed or modified after the change.
•Each cache list has an additional weighting factor that is used to compute a score for the cache
block list. The score for a particular list is the weighting factor multiplied by the access time of
the oldest cache block on the list.
•A special weighting factor can be assigned specifying that all other cache lists should be
searched before this list is searched. This stops the oldest cache block age from being used to
create a score for a given cache list.
•The weighting factor can be changed dynamically and the new settings immediately take effect.
•If the weighting factor is assigned to multiple cache lists, an implied priority is established based
on the cache list number (in other words, search the cache list with the lower priority level first).
•The special weight factor for list 1 CANNOT be set to zero. Initially, all of the cache blocks
available to the system live in the first list and setting the weight factor on this list causes
undesired effects. Setting the weight to zero is interpreted by the system as ìonly use this list if
no other cache blocks in the system are available.î The moment a file has been written to media
and the cache blocks are put back on the free list, the blocks will be re-used immediately, even
if there are thousands of unused cache blocks available on list one (with a weight of zero).
•A cache block is reallocated from the list with the highest score for all eight cache priority lists.
•All cache blocks exist on one list. Cache block reallocation occurs on a first-in-first-out basis.
•medialist—Users with large volumes can limit the medialist output to show only drive status.
•-j Directs the medialist command to jukebox ID rather than the default of 1.
•Network Attached Libraries—Enhancements to the network attached library interface include:
•Reducing the number of 'query drive' commands thus softening the load on the XDI interface.
A function to test the need to inquire about library drive status is added. If there are no drive
candidates to service a pending IO request, the library query drive command is not called. This
will reduce the load on the network library interface.
•Releasing the shared memory semaphore across the library query drive command. Failure to
do this will cause the entire AMASS system to lock up should a query drive command fail to
complete (is hung).
•Adding a feature that will time out a hung XDI command. This happens frequently in the
presence of library errors.
•Adding XDI failed mount/dismount retry option. XDI failed mounts and dismounts are retried 3
times before a drive/volume are taken out of service.
The combination of these four fixes will prevent AMASS from locking up and loosing drive/volume
resource in the presence of library errors.
This new functionality is controlled by using the following environmental variables:
•AMASS_DISABLE_TEST_DRIVE - should the reduction in queries to the drive cause undesired
effects, this environmental variable will return AMASS to pre-fix default state that continually
pings the library interface.
•AMASS_DRIVESTAT_TIMEOUT=<new_val_in_seconds> - changes query drive time out
from the default of 10 seconds to the value specified here.
•AMASS_STORE_TIMEOUT=<new_val_in_seconds> - changes the library mount/dismount
timeout from a default 360 seconds to the value specified here.
•AMASS_DISABLE_LIBTIMEOUT - should the library timeout feature cause undesired effects,
this environment variable will disable the feature.
Note
•vgimport—Importing and exporting large volume groups is expedited with the -q option.
Adds the value vlk_maxxfr found in amass/scripts/amassconfig2 to the volume
capacity.
•-F Sets a tape volume to FULL.
•-s Sets volume available (AVAIL) size (Mb)
volstat -s <available size in Mb> <vol-num>
•-u Displays this usage message.
•-y Suppresses confirmation and informational messages.
If -a or -i are not specified, then the active status of the specified volume is toggled.
Additional Support
Additional operating system support includes:
•AIX 5.2 (32-bit)
•HP Tru64 5.1B
•SGI IRIX 6.5.21m
•SGI IRIX 6.5.22m
•SGI IRIX 6.5.23
Additional drive support includes:
•AIT-3 WORM
•HP Ultra Density Optical (UDO)—optical format and in selected ux-series libraries
•IBM 3592
•Plasmon Ultra Density Optical (UDO)—optical format and in selected G-series libraries
•SDLT 320
•STK 9840B
Additional library support includes:
•ADIC Scalar 10K in a Dual Aisle configuration
The current version of AMASS does not support the high availability and fail over feature sets of
the Scalar 10K in a Dual Aisle configuration. AMASS does support the Scalar 10K in a Dual Aisle
configuration as a large 10K library.
The Plasmon G-Series libraries were previously supported with optical drives.
Mixing drive types within the library is not currently supported.
•G64
•G104
•G164
•G238
•G438
•G638
Additional firmware support includes:
•ACSLS 7.0
•ACSLS 7.1
•Scalar i2000 library M1 firmware
•Scalar i2000 library M2 firmware
•Scalar DLC 2.5 Service Pack 1
Software Installation
Use the information provided in this section to upgrade your AMASS for UNIX to Version 5.4.
System Requirements
The requirements identified here are necessary to support the installation of AMASS for UNIX Version 5.4:
System/ComponentRequirement
Operating System• The operating system must always be run in US English.
• Your native operating system, and not AMASS, limits the maximum size
of your files.
• AMASS supports only the maintenance (m) and not the feature (f) stream
of IRIX 6.5.x.
• For operating systems in which both 32-bit and 64-bit versions of AMASS
exist, the bit version of the AMASS API must match the bit version of the
AMASS applications.
Server PlatformTo obtain details on supported application server platforms, contact your
AMASS sales representative.
Required Operating System Patch Levels
ADIC requires operating system patches to successfully operate AMASS. Obtain these patches from the
appropriate vendor. To view a list of patches that are already installed on your machine, go to View Patches.