Quantum STORNEXT FX 2.0 User Manual

StorNext FX 2.0 Release Notes

Contents 0

Purpose of this Release ................................................................................................. 2
Operating System Level Requirements ...................................................................... 2
Resolved Issues .............................................................................................................. 8
Known Issues ............................................................................................................... 11
Operating Guidelines and Limitations..................................................................... 12
Documentation............................................................................................................. 18
Contacting Quantum................................................................................................... 19
Document 6-01663-03 Rev A
Quantum Corporation provides this publication “as is” without warranty of any kind, either express or implied, including but not limited to the implied warranties of merchantability or fitness for a particular purpose. Quantum Corporation may revise this publication from time to time without notice.
COPYRIGHT STATEMENT
Copyright 2008 by Quantum Corporation. All rights reserved.
Your right to copy this document is limited by copyright law. Making copies or adaptations without prior written authorization of Quantum Corporation is prohibited by law and constitutes a punishable violation of the law.
DLT and DLTtape are trademarks of Quantum Corporation. Quantum, the Quantum logo, and the DLTtape logo are all registered trademarks of Quantum Corporation. Other trademarks may be mentioned herein which belong to other companies.
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008

Purpose of this Release 0

StorNext FX (SNFX) 2.0 release notes describe known issues and issues that were resolved for this release. These notes also provide drive and auxiliary firmware compatibility information.
StorNext FX is a client only version software that is licensed to run in Apple managed environments and used with an Apple metadata controller. It is fully interoperable and compatible with Apple Xsan. StorNext FX 2.0 is built using StorNext File System components from the StorNext 3.1 release.
Visit www.quantum.com/ServiceandSupport for additional information and updates for StorNext FX.

Operating System Level Requirements 0

Support for the following operating systems has been added in StorNext FX 2.0:
•Red Hat® Enterprise Linux® 5
•SUSETM Linux Enterprise Server 10
Starting with StorNext FX 2.0, the following operating systems are no longer supported:
Windows 2000
Red Hat Enterprise Linux 3
•AIX 5.2
SuSE Linux Enterprise Server 9
Note: For systems running Red Hat Enterprise Linux version 4 or 5, before
installing StorNext you must first install the kernel header files (shipped as the kernel-devel-smp or kernel-devel RPM).
For systems running SUSE Linux Enterprise Server, you must first install the kernel source code (typically shipped as the kernel-source RPM).
®
Purpose of this Release 2
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Note: GNU tar is required on Solaris systems. In addition, for systems
running Solaris 10, install the Recommended Patch Cluster (dated March 10, 2006 or later) before installing StorNext.
To enable support for LUNs greater than 2TB on Solaris 10, the following patches are required:
• 118822-23 (or greater) Kernel Patch
• 118996-03 (or greater) Format Patch
• 119374-07 (or greater) SD and SSD Patch
• 120998-01 (or greater) SD Headers Patch
Caution: Red Hat 5 ships with secure Linux kernel enabled by default. To
ensure proper StorNext operation, you must not install Red Hat 5 with secure Linux enabled. The secure Linux kernel must be off, or the file system could fail to start.

StorNext FX 2.0 Requirements

The following table lists the required operating system levels required to successfully operate StorNext FX 2.0
0
Operating System Platform Operating System Levels
Red Hat Linux Red Hat Enterprise Linux Advanced
Server 4.0 - Updates 4 and 5 (32-bit Intel and AMD; 64-bit Itanium, Opteron, and EM64T)
Required kernels:
2.6.9-42.EL (Update 4)
2.6.9-55.EL (Update 5)
Red Hat Linux Red Hat Enterprise Linux Advanced
Server 5.0 (64-bit Opteron, and EM64T)
Required kernels:
2.6.18-8
Operating System Level Requirements 3
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
Operating System Platform Op erating System Levels
April 2008
Sun Solaris
TM
SolarisTM 9 and 10 (SPARC 64-bit)
Generic_118855-14 release
NOTE: Prior to installing StorNext on a Solaris 10 machine, you must install the Solaris 10 Recommended Patch Cluster dated March 10, 2006 or later.
NOTE: To gain support for LUNs greater than 2TB on Solaris 10, the following patches are required:
118822-23 (or greater) Kernel Patch
118996-03 (or greater) Format Patch
119374-07 (or greater) SD and SSD Patch
120998-01 (or greater) SD Headers Patch
SuSE Linux SUSE Linux Enterprise Server 10,
Service Pack 1; (32-bit Intel and AMD, 64-bit Itanium, Opteron, and EM64T).
Required kernels
2.6.16.21-0.8-default
2.6.16-46-0.12 (SP1)
Both the kernel and kernel source RPM packages must be installed. Install all tools necessary to build a kernel module (including compilers) on all Linux clients and server systems.
Windows Server 2003
®
Service Pack 2 (32-bit and 64-bit Intel and AMD)
Windows XP® (Client only) Service Pack 2 (32-bit and 64-bit Intel
and AMD)
Windows Vista
IBM AIX
HP-UX
®
SGI IRIX
®
®
(32-bit and 64-bit Intel and AMD)
5.3 (Power Architecture 64-bit)
11i v2 (PA-RISC and 64-bit Itanium)
®
6.5.29 and 6.5.30 (MIPS 64-bit)
Operating System Level Requirements 4
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008

Certified System Components

The following table lists certified system components that have been Quantum tested and qualified to support StorNext FX 2.0. For details on other supported
0
components, contact the Quantum Technical Assistance Center
Component Description
Browsers Netscape 7.x
Microsoft Internet Explorer 5.5 and later Mozilla 1.4 and later FireFox 1.0 and later Minimum browser resolution: 800x600
HBA Emulex LP8000, LP850, LP9000,
LP90002 QLogic: QLA2200, QLA2310, QLA2340, QLA2342
NFS Versions 3
Multipathing Failover Software Red Hat Enterprise Linux AS 3.0,
Update 3 HP SecurePath - version: 3.0c
Windows 2003 EMC PowerPath - version: 3.0.6
For information on supported HBA drivers, refer to the RAID vendor’s documentation.

System Requirements 0 This section describes the system requirements necessary to successfully install

StorNext FX 2.0.
Note: The following requirements assume that StorNext FX is the only
application running on your system.
Operating System Level Requirements 5
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
StorNext FX 2.0 File System Connectivity Requirements 0
The following connectivity requirements must be met for each StorNext FX mounted file system:
System/Component Requirement

StorNext Client Software Requirements

Network LAN using TCP/IP (all clients and servers must be interconnected)
For SNFX metadata traffic, Quantum requires that a separate, dedicated, switched Ethernet LAN be used.
SAN SNFX clients: An FC-HBA or
equivalent SAN communication device where the storage is visible and accessible to multiple SAN clients.
SNFX does not support multiple hosts connected through an FC hub device because the resulting propagation of Loop Initialization Protocol resets can cause data corruption.
To install and run the StorNext client software, the client system must meet the following minimum hardware requirements.
0
For SAN (FC-attached) clients:
•1 GB RAM
500 MB available hard disk space

StorNext File System Client Buffer Cache

For StorNext FX 2.0, multiple mounted file systems typically share a single buffer cache. A StorNext client buffer cache is created for each different
0
default, all file systems have the same
cachebufsize of 64K, so they all share the
cachebufsize. By
same buffer cache.
Note: These settings do not apply to MacOS X clients, which do not use the
StorNext buffer cache.
The amount of memory consumed by default for each cachebufsize depends on the platform type and the amount of memory in the system. The following table shows the default amount of memory consumed by
cachebufsize:
Platform Type <=2GB Memory >2GB Memory
32-bit Windows 32MB 64MB
Operating System Level Requirements 6
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Platform Type <=2GB Memory >2GB Memory
32-bit Linux 64MB 128MB
All other platforms 64MB 256MB
To see information about the buffer cache after mounting file systems, use the
cvdb(1) command with the -b option. To change the amount of memory used by
the buffer cache at mount time, use the
buffercachecap parameter.
On Windows, the non-paged pool is used for buffer cache memory until it consumes up to 64 megabytes (32-bit systems) or 64 gigabytes (64-bit systems). Any additional buffer cache memory comes from the paged pool.

The Reserved Space Parameter

The method of accounting for reserved space has changed for StorNext FX 2.0. The
0
MaxMBPerClientReserve parameter from the StorNext file system
configuration file (
/usr/cvfs/config/*.cfg) has been deprecated. All values except 0
are ignored for this parameter. In addition, there is a new parameter,
ReservedSpace.
The ReservedSpace parameter lets the administrator control the use of delayed allocations on clients.
ReservedSpace is a performance feature that lets clients
perform buffered writes on a file without first obtaining real allocations from the metadata controller (MDC).
The ReservedSpace parameter can be set to Yes or No:
Yes - (Default) The MDC reserves enough disk space so that delayed
allocations can be converted to real allocations (even when the MDC is restarted and the client is not). The MDC reserves a minimum of about 4GB for each stripe group and up to 280MBs per actively writing client for each stripe group.
Note: The amount of reserved space is usually less than 280MBs per
client. Reserved space is calculated as 110% of the buffer cache size of each particular client. For example, a client with a 64MB buffer cache is allocated 70MBs of reserved space by the MDC. If the client closes all files that are open for write, the 70MBs of space is no longer accounted for. It is important to remember that
reserved space is per stripe group.
No - More disk space is available for use, but buffer cache performance is
affected, and fragmentation may occur.
If the MaxMBPerClientReserve parameter exists in the configuration file and has a value of
0, ReservedSpace is set to No. Otherwise, ReservedSpace defaults to Yes.
Operating System Level Requirements 7
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Note: In prior releases of StorNext FX, MaxMBPerClientReserve defaulted to
100MBs, and reserved space was the product of multiplied by the actual amount of reserved space that clients use but caps the value to about 280MBs per client.
In addition, the MDC maintains a “minimum level” of reserved space. As a result, in some cases, more reserved space may be visible. Reserved space appears as allocated disk space per data stripe group.
The minimum reserved space is enough to accommodate 15 clients or
MaxConnections – 1, whichever is lower. For example, if a cluster has a MaxConnections of 3, the reserved space total can be under 1GB.
MaxConnections – 1. In StorNext FX 2.0, the MDC tracks
MaxMBPerClientReserve

Resolved Issues 0

The following table lists resolved issues that are specific to StorNext File System and might therefore affect StorNext FX.
Note: There is no change to cryptographic functionality in StorNext FX
release 2.0.
Operating System CR Number SR Number Description
Linux 16616 n/a Resolved a condition in which a file system could not
be created after upgrading from SuSE 9 to SuSE 10 due to an issue with the ksh.
16627 n/a Removed erroneous error messages that appeared
after a successful client installation on a SUSE 10 machine.
16697 n/a When “cp -p” is run from a Linux NFS client
connected to a Solaris NFS server that exports a StorNext file system, the following error message could appear:
linux# cp -p foo bar cp: preserving permissions for `bar': Input/output error
23299 n/a Corrected a condition which caused the FSM brl
resync to fail if FSM was bounced too soon after failover.
HP-UX 16625 n/a Corrected a condition in which SNFS on HP/UX
didn't recognize LUN numbers greater than 8.
Resolved Issues 8
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System CR Number SR Number Descriptio n
Solaris 16697 n/a When “cp -p” is run from a Linux NFS client
connected to a Solaris NFS server that exports a StorNext file system, the following error message could appear:
linux# cp -p foo bar cp: preserving permissions for `bar': Input/output error
Windows 16222 n/a Resolved an error condition in Proxy Server Packet
Handling (ProcessPkt).
16379 600546 Safeguards added to catch > PASSIVE_LEVEL
requests on fast path and force using irp.
16661 n/a Corrected a condition in which SNFS allowed writes
from the Window 2003 LDM.
16680, 16681 and 23300
621732 Corrected a condition in which Alloc_space didn't
round appropriately for perfectfit nomorethan
allocations.
23031 n/a Resolved a condition in which cvupdatefs -R didn't
work properly on Windows machines.
23071 n/a Addressed an issue that caused FSM to crash instead
of exiting when it encountered an error.
23142 n/a Corrected a condition that generated an error message
when using the example.cfg file for the “Modify
Existing” function.
23144 722239 Corrected a condition in which CvReadDir entered an
infinite loop.
23146 n/a Corrected a condition in which an “Error: Internal
Error on Allocation Strategy” appeared after changing file system configuration stripe group parameters and rebooting the machine.
23165 719275 Resolved a condition in which a file system could not
be mapped when there were more than 26 available SNFS file systems, but the maximum number of file systems was not mounted. (You cannot mount more than 26 file systems.)
23231 724263 Corrected a condition that caused the last block of files
in a directory to be omitted when a perl script was used to populate a directory.
23233 724154 Corrected a condition in which a file could be deleted
from a directory on SNFS with a given ACLs setup, but NTFS did not allow a file to be deleted with the exact same ACLs set.
23252 n/a Resolved a condition in which SNFS directories
appeared as mounted volumes in Windows Explorer.
Resolved Issues 9
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System CR Number SR Number Descriptio n
Windows 23288 n/a Corrected a condition in which running the rmsvc
command on a Windows machine could cause the following to happen:
• The FSMPM and all local FSMs are shut down
• The client and filter continue to run, and all file systems remain mounted and mapped
• Any requests to access the StorNext file system could hang
All requests bound for local FSMs could hang (unless failover is running)
23292 708870 Corrected a condition in which a file on a StorNext file
system accessed from another machine via CIFS didn’t update the size on the CIFS client.
23320 n/a Resolved a condition that locked up the system after
starting FSM services.
All 15910 n/a Added improvements to the cvdbset script.
16012 n/a Reduced the size of the free space tree structures.
16019 717096 Corrected a condition in which starting FSM while
backups are running caused the restore journal to be disabled.
16040 n/a Resolved a condition in which data corruption
occurred when device mappings from a SAN environment changed while CVFS is operational.
16041 n/a Reduced FSM CPU consumption in transaction buffer
processing.
16078 n/a Corrected a situation in which cvmkfs didn't catch an
error when devices are rearranged underneath it.
16094 n/a Corrected an error related to XSan 2.0 which caused
client core dumping during a mount attempt.
16142 n/a Addressed an FSM ASSERT failure.
16378 710517 Corrected a condition in which mounting failed due to
the kernel misinterpreting the number of valid paths.
16664 712214 Corrected an error allocating new directory blocks
due to copious copying and moving files to a different directory.
23083 n/a Corrected a condition in which running the proxy who
command caused the fsmpm process to spin endlessly on the target machine, repeatedly asking the kernel for the proxy client list.
Resolved Issues 10
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System CR Number SR Number Descriptio n
All 23122 n/a Addressed a condition in which it was not possible to
stop an FSM if the IP addresses change. Running the
cvadmin stop command caused failure with the error
message “FSM is not local.”
23143, 23302 and 23309
23230, 23323 and 23339
23411 and 16587
720143 Resolved directory cache invalidation errors when
dragging and dropping files from a Mac to a Windows machine.
725383 Added support for the DS4700 array with inquiry
string “IBM 1814” as a known “LSI” type disk.
715497 Corrected the error message that appears if opening
/dev/cvfsctl fails with EPERM

Known Issues 0

The following table lists known issues in this release of StorNext FX, as well as associated workarounds where applicable.
Operating System
Macintosh 24577 n/a Adding a Macintosh client to an
CR Number
SR Number
Description Workaround (if applicable)
After you add a Macintosh existing Xsan cluster and then rebooting the metadata controller
causes a change in the .auth_secret
file on all Macintosh systems in the cluster. Unless you modify the
.auth_secret file on your SNFX client,
errors could result.
client to an existing Xsan
cluster, you must copy the
.auth_secret file from the
Macintosh systems to SNFX
clients.
Known Issues 1 1
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System
All
CR Number
23474
SR Number Description Workaround (if applicable)
n/a Due to the way SNFS performs write-
behind caching for buffered I/O and flushes buffers to disk, you might receive some of the following system log entries, depending on your system configuration:
I/O error on cookie 0x36a11a cvfs error 'IO error' (0x3) os status code 0x3 blkno 0xaad9c000 offset 0x5a40000 count 0x4000 resid 0x0
I/O path in error [hostid: 5 lun: 0 path < /dev/sdb>] quiesced for 300 seconds
I/O error on cookie 0x36a11a cvfs error 'IO error' (0x3) os status code 0x3 blkno 0xaad9c000 offset 0x5a40000 count 0x4000 resid 0x0
When these messages appear,
you should check the system
log for driver-level errors that
generally indicate data loss
due to SAN hardware
problems.
Note: When random I/O is
performed, there is also a rare
case that can generate a
spurious I/O error with a
status code of 0x2. For
example:
I/O error on cookie 0x36a11a
cvfs error 'Not found' (0x2) os
status code 0x2 blkno
0xaad9c000 offset
0x5a40000 count 0x4000
resid 0x0
These errors can be safely
ignored.

Operating Guidelines and Limitations 0

The following table lists updated information and guidelines for running StorNext FX, as well as known limitations in this release. (Some of the guidelines and limitations listed apply to previous versions of StorNext upon which StorNext 2.0 is based.)
Operating System / Affected Component
IRIX StorNext supports only SGI 64-bit versions of IRIX on operating
Description
systems that use the SGI version of the QLogic QLA2200 or QLA2310 FC-HBAs.
SNFS uses 64-bit inode numbers. 32-bit applications (programs compiled without 64-bit) support may experience problems working with files with inode numbers greater than 2147483648. 32-bit applications may also encounter problems when dealing with files greater than 2TB in size, This issue is not unique to StorNext, but may be encountered by some legacy applications when working with SNFS on IRIX.
Operating Guidelines and Limitations 12
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System / Affected Component Description
IRIX On many versions of IRIX, the root crontab contains the following entry
which is used to remove old application crash dumps and temporary mail files:
find / -local -type f '(' -name core -o -name dead.letter ')' -atime +7 -mtime +7 ­exec rm -f '{}' ';'
If StorNext file systems are mounted, they will be traversed by this find command which can have a dramatic impact on the performance of other applications currently using these file systems. To prevent the traversal of StorNext file systems, modify the
find / -local '(' -type d -fstype cvfs -prune ')' -o -type f '(' -name core -o -name dead.letter ')' -atime +7 -mtime +7 -exec rm -f '{}' ';'
find command so it reads:
Linux StorNext does not support the use of loopback IP addresses other than
127.0.0.1. SUSE Enterprise Linux 10 systems sometimes have an entry for 127.0.0.2 in the
/etc/hosts file. This causes various FSM operations
to fail with the error “Not Local.”
Linux Many versions of Linux run a cron script nightly to build a database
used by the they are traversed by this
slocate command. If StorNext file systems are mounted,
cron job which can have a dramatic impact
on the performance of other applications currently using these file systems. Perform the following steps (based on Linux version) to prevent the
RedHat Enterprise Linux 3
cron script from traversing StorNext file systems.
Add “cvfs” to the list of file system types to be skipped. This is usually done by modifying the “updatedb” command line in the
cron.daily/slocate.cron
/usr/bin/updatedb -f “cvfs,nfs,smbfs,ncpfs,proc,devpts” -e “/tmp,/var/tmp,/usr/ tmp,/afs,/net”
file to read:
/etc/
Also, modify the /etc/updatedb.conf file by adding “cvfs” to the line beginning with “PRUNEFS” so that it reads:
PRUNEFS=”cvfs devpts NFS nfs afs sfs proc smbfs autofs auto iso9660”
RedHat Enterprise Linux 4
Add “cvfs” to the list of file system types to be skipped. This is usually done by modifying the “PRUNEFS” line in the
/etc/updatedb.conf file
to read:
PRUNEFS=”cvfs sysfs selinuxfs usbdevfs devpts NFS nfs nfs4 afs sfs proc smbfs cifs autofs auto iso9660 udf”
SUSE Linux Enterprise Server 9 and 10
The optional “findutils-locate” package is used to build the slocate database. The default behavior is to disable building the database. If enabled, to prevent cvfs file systems from being scanned, add “cvfs” to the list of file system types to be skipped. This is usually done by modifying the “UPDATEDB_PRUNEFS” line in the
/etc/sysconfig/
locate file to read:
UPDATEDB_PRUNEFS=”cvfs”
Operating Guidelines and Limitations 13
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System / Affected Component Description
Solaris On Solaris, by default, the /usr/lib/fs/nfs/nfsfind script is run nightly by
the
cron daemon. This script contains a find command that traverses
any local file system that is exported (shared) via NFS. If the running of this script is interfering with the performance of processes accessing StorNext file systems, it can be modified to skip them. To do so, add the following line to the
'(' -type d -fstype cvfs -prune ')' -o
find command in the script:
The final find command should look like this:
find $dir '(' -type d -fstype cvfs -prune ')' -o -type f -name .nfs\* -mtime +7 ­mount -exec rm -f {} \;
Solaris The Solaris Security Toolkit, formally known as JASS, causes the
following two issues:
• It disables RPC by renaming the RPC startup script, disrupting the
StorNext interprocess communication. To fix the communication problem, rename the RPC startup script in
extension>
to rpc.
/etc/init.d from rpc.<illegal
• It turns on IPSec, causing numerous warning messages in the
system log file. Either disable IPSec by removing the IPSec startup file in
/etc/init.d or contact Sun Technical Support to find out how to
reconfigure IPSec to ignore local loopback connections.
All Unix and Linux The swapon command does not work on StorNext file systems. The
Linux/Unix paging and swapping take place. If
swapon command is used to specify devices on which
swapon is run on a StorNext file
system, the command fails with an invalid argument error. In a file system configuration file, if the Metadata parameter for a stripe
group was set to change the parameter to
Yes when the file system was created, do not later
No. Doing so will cause the metadata to
become inaccessible and result in errors. Also, if the parameter is changed to
No, the cvupdatefs command might allow bandwidth
expansion of the stripe group, resulting in metadata corruption.
Operating Guidelines and Limitations 14
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
Operating System / Affected Component Description
Windows Windows Services for UNIX (SFU) supports only NTFS for NFS
exports. Because of this limitation, a Windows system cannot act as an NFS server for StorNext File System.
On Windows systems, StorNext may log error messages after reboot if it tries to start before some other services. For example, if the network is not yet available, a StorNext client cannot contact a StorNext server and posts an error in the logs. (The client will continue to retry the connection until it is successful.)
To prevent this from occurring, make the startup of StorNext services dependant on another Windows service (for example, the Workstation service). For more information, see the Microsoft Knowledge Base article Q193888, “How to delay loading of specific services.”
Windows-based SNFS clients do not support symbolic links that point to a file system outside of SNFS. Symbolic links are created either by a UNIX/Linux client or by a Windows client (and are referred to as junctions).
April 2008
Virus-checking software can severely degrade the performance of any file system, including SNFS. If you have anti-virus software running on a Windows Server 2003 or Windows XP machine, Quantum recommends configuring the software so that it does NOT check SNFS.
Windows If you are using the StorNext client software with Windows Server
2003 or Windows XP, turn off the Recycle Bin in the StorNext file systems mapped on the Windows machine, so the file systems will work properly.
1 On the Windows client machine, right-click the Recycle Bin icon
on the desktop and click Properties.
2 Click Global. 3 Click Configure drives independently. 4 Click the Local Disk tab that corresponds to the mapped file
system.
5 Select the Do not move files to the Recycle Bin. Remove files
immediately when deleted check box.
6 Click Apply, and then click OK.
Operating Guidelines and Limitations 15
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System / Affected Component Description
All In StorNext 3.0, the default buffer cache settings have been modified.
Previously, all reads/writes that were 64K or smaller went through the buffer cache while larger I/O requests went direct. In StorNext 3.0, read/writes that are 1MB or smaller go through the buffer cache, while larger I/O requests go direct.
The new buffer cache settings may change the I/O behavior of some applications. For example, on managed servers, I/O to and from tape now goes through the buffer cache. To revert to the settings used in previous releases, change the following mount options on StorNext clients:
auto_dma_read_length=65537 auto_dma_write_length=65537
Hot re-zoning of SAN fabrics is not supported.
StorNext File System is incompatible with third-party portmappers. In StorNext 3.0, the buffercachemin and buffercachemax parameters are
deprecated and are replaced by
buffercachecap.
Also, in the Client Configuration Windows utility, the Buffer Cache Min and Buffer Cache Max settings are replaced by Buffer Cache Cap. The Non-Paged Pool Percentage setting has been removed.
Operating Guidelines and Limitations 16
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008
Operating System / Affected Component Description
All In StorNext 3.0, obsolete methods of specifying FSS addresses are no
longer supported. If the fsroutes configuration file is present, it is ignored and a warning
is generated. In addition, the FSM address field for each entry in the
fsmlist configuration file (the second column) is no longer supported
and is ignored if present. If the priority field (the third column) is specified, then a period (.) must be used to take the place of the second column.
The correct way to specify the metadata network for a cluster is through the addresses in the address of a name server in network
<x> will be used as a metadata network.
fsnameservers configuration file. If the
fsnameservers is on network <x>, then
As of SNFS 2.7, a change was made to the way that the Reserved Extents performance feature affects free space reporting. In the previous release, SNFS would reserve a certain amount of disk space which would cause applications to receive an
out of space error before
the disk capacity reached 100%.
In the current release, this reserved space is treated as allocated space. This allows applications to perform allocations until the file system is nearly full.
NOTE: Due to allocation rounding, applications may still receive a
premature
out of space error, but only when there are just a few
megabytes of space remaining. In the worst case, the error will be returned when the reported remaining space is:
(InodeExpandMax * #-of-data-stripe-groups)
One side effect of this change is that after creating a new file system, df will show that space has been used, even though no user data has been allocated.
The amount of reserved space varies according to client use but does not go below a “floor” of a few gigabytes per data stripe group. The amount of reserved space at any time can be seen using the
cvadmin
command, selecting the file system, and using show long.
While not recommended, the Reserved Extents feature can be disabled by applying the following setting to the Globals section of the FSM configuration file:
ReservedSpace No
This will cause the file system to not reserve space for buffered I/O, thereby reducing buffer cache performance and possibly causing severe fragmentation.
For more information, see the cvfs_config(4) man page.
Operating Guidelines and Limitations 17
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008

Documentation 0

The following documents are currently available for StorNext products:
Document Number Docum ent Title
6-01658-01 StorNext User’s Guide
6-00360-10 StorNext Installation Guide
6-01376-05 StorNext File System Tuning Guide
6-01743-02 StorNext FX File System Quick Reference Guide
To order a copy of the source code covered by the Open Source License Agreement, contact the Quantum Technical Assistance Center.
Documentation 18
StorNext FX 2.0 Release Notes
Document 6-01663-03 Rev A
April 2008

Contacting Quantum 0

More information about this product is available on the Quantum Service and Support website at Service and Support website contains a collection of information, including answers to frequently asked questions (FAQs). You can also access software, firmware, and drivers through this site.
To request a software upgrade, visit www.quantum.com/ServiceandSupport/
Upgrade/Index.aspx. For further assistance, or if training is desired, contact
Quantum Global Services:
www.quantum.com/ServiceandSupport. The Quantum
North America
UK, France, and Germany
EMEA
World Wide Web
+1 800-284-5101
00800 4 QUANTUM
+44 1256 848 766
www.quantum.com/ServiceandSupport
Contacting Quantum 19
Loading...