- 1 IRIX 6.5.15 Feature Stream:
- 2 IRIX 6.5.15 Maintenance and Feature Stream:
- 3 Next release
- 4 Previous release
- 5 See also
IRIX 6.5.15 Feature Stream:
XVM Subvolume naming convention
The default naming convention for subvolumes in the /dev/lxvm/ and /dev/cxvm directories is volname, subvolume. Group quotas for XFS filesystems Disk quotas maybe implemented on XFS filesystems according to group ID. Previous releases of IRIX supported implementing quotas according to user and project ID. Project and group accounting are mutually exclusive. Caution: Group quotas are supported in the feature stream only. Do not implement group quotas on a disk and, subsequently, mount that disk with the pquota mount option on a machine running the maintenance stream.
CXFS new features
Default scripts are now provided in the /var/cluster/clconfd-scripts directory to permit NFS-exporting of CXFS filesystems listed in /etc/exports. Changes have been made to the rotatelogs script syntax. The root crontab file now has an entry to run the rotatelogs script weekly. If you run the script twice in one day, it will append the current log file
CXFS upgrade: filesystem definitions
The structure of the CXFS filesystem configuration was changed with the release of IRIX 6.5.13f. Backward compatibility with earlier versions is no longer maintained as of IRIX 6.5.14f, since all nodes in the cluster must be running the same or adjacent releases. If you are upgrading from 6.5.13f to 6.5.14f or later, there is no further impact. If you are upgrading from 6.5.12f or earlier, you must perform a one-time manual conversion of your CXFS filesystem definitions.
CXFS upgrade: cluster nodes
For CXFS customers, if you install IRIX 6.5.15f including sysadm_base on a cluster node where the the 6.5.13f or earlier version of CXFS is installed, you will need to upgrade to the 6.5.15f version CXFS (sysadm_cxfs) and install 6.5.15f sysadm_cluster.
IRIX 6.5.15 Maintenance and Feature Stream:
XFS inode numbers are 64-bit values containing an encoded disk location. For large filesystems (1 Terabyte plus), the inode numbers can overflow into the top 32 bits of the number. Certain backup applications which use the inode number only allow 32 bits of space, and thus have severe problems dealing with very large XFS filesystems. In IRIX 6.5.15, XFS has been changed to allocate inodes only within the lower portion of the filesystem to avoid this overflow issue. Other allocation policies in the filesystem will make this lower order space preferred for inodes and reduce the risk of a filesystem refusing to create new inodes when it still has space available.
For existing large filesystems with high numbered inodes this does not solve the problem, but it does solve it for new filesystems, and for new files created on existing filesystems.
For backwards compatibility a new mount option is available, inode64, this will allow XFS to place inodes anywhere in the filesystem and follow exactly the original placement policies. Filesystems which were small enough not to be able to overflow 32 bits of inode number also retain the old behavior.
The 6.5.15 release of IRIX adds support for disks that have a capacity of 2 terabytes. Previously, the maximum was 1 terabyte. The maximum disk capacity had become an issue with the introduction of 180GB drives, multiples of which can be combined into a single logical unit (lun) behind a RAID controller. The maximum capacity of a logical unit for both of our current RAID controllers is 2 terabytes, so this change allows us to support that maximum.
Documentation of chproj(1) Chapter 5 "System Administration in a Multiuser Environment" in IRIX Admin: System Configuration and Operation has been updated with a new section on Project ID numbers and the chproj(1) command that changes the project ownership of a file. Default value of gang_sched_off The systune parameter gang_sched_off is turned on by default. This means that gangs will not be running on the system unless the system administrator sets gang_sched_off to 0. This differs from previous IRIX releases, for which gangs ran on the system automatically. This change has been made per field request because we have discovered that having gang scheduling on by default greatly diminishes system performance. Most customers have been running with it off.
A coremask parameter is set in the /var/sysgen/mtune/kernel file to determine what kind of permissions a core file will receive. The old behavior uses the umask of the owner of the process to determine the permissions. New partmgr(1M) command A new partitioning administration command, partmgr(1M), is available to define and store partition definitions.
/dev/poll device added Support has been added for the /dev/poll interface for scalable and efficient event notification. DHCP Setup The Dynamic Host Configuration With Proclaim section in IRIX Admin: Networking and Mail has been updated to include detailed procedures and descriptions for setting up DHCP. New DHCP client options The DHCP client (proclaim) has been updated to include options to allow DHCP clients to capture changes to DNS, NIS, or GATEWAY in your environment. System Availability
For FailSafe customers, if you install IRIX 6.5.15, including sysadm_base, on a cluster node where FailSafe 2.1.1 is installed, you will need to upgrade to FailSafe 2.1.2 (sysadm_failsafe2) and install 6.5.15 sysadm_cluster.
Designating CPUs in the cpuset configuration file This release provides the ability to designate one or more CPUs or a range of CPUs on a single line in the cpuset configuration file. The CPUs in a cpuset configuration file do not have to be specified in a particular order.
System Performance and Tuning
Update to systune(1M) man page The systune(1M) man page has been updated to describe a situation where the system segment size value (syssegsz) reported by the systune command is different than the value set in the /unix file.
Origin 300 Systems
SGI Origin 3000 server (Onyx 3000 server)
Partitioned FLEXlm licensing change Starting with the IRIX 6.5.15 release, each partition of a partitioned Origin 3000 system will have a new unique FLEXlm host ID (lmhostid). Any FLEXlm license based on the old non-partitioned lmhostid value will no longer work and a new license key must be generated. See the following web page for additional details, /support/licensing/. Contact the applicable software product vendor and provide the lmhostid number of the partition to generate a new license key. For nodelock licenses, provide the lmhostid for all partitions of the software installed and operating. For software licensed by SGI, see the following web page for more details: /support/licensing/.
lmparthostids and lmbasehostid commands Support for the lmparthostids and lmbasehostid commands has been added. The lmparthostids command generates a list of all 63 partition specific hostids and the unpartitioned id. The lmbasehostid command takes any arbitrary lmhostid and generates the base unpartitioned id. Updated FLEXlm v7.2i support from GLOBEtrotter Software, Inc. The license manager daemon and the floating license server are built in V5 compatibility mode. For backwards compatibility, the client libraries and header files are still based on FLEXlm v6.1g.