C H A P T E R

C H A P T E R 1

SSP 3.5 Release Notes

This chapter covers the following topics:

General Issues

This section discusses general issues for SSP 3.5 software.

Alternate Pathing Functionality

If you are currently using alternate pathing (AP) software on your Sun Enterprise 10000 domains with dynamic reconfiguration (DR), note that you must use multipathing software compatible with the type of DR model (either 2.0 or 3.0) running on those domains. For further information on DR models, see Changes to the SSP Software .

For multipathing on DR model 2.0 domains, run the Sun Enterprise Server Alternate Pathing software. For further AP information, refer to the Sun Enterprise Server Alternate Pathing User Guide.
For multipathing on DR model 3.0 domains, run Internet Protocol Multipathing (IPMP) software provided with the Solaris 8 or 9 operating environment--and the Sun StorEdge Traffic Manager software (formerly known as Multiplexed Input and Output (MPxIO) software). For details on IPMP software, refer to the IP Network Multipathing Administration Guide available in the Solaris 8 operating environment or the System Administration Guide: IP Services documentation available in the Solaris 9 operating environment.

For details on the Sun StorEdge Traffic Manager, refer to the Sun StorEdge Traffic Manager Software Installation and Configuration Guide available on the Sun Network Storage Solutions Web site ( http://www.sun.com/storage/san ).

Note - The Sun StorEdge Traffic Manager software may not support automatic path switching for all devices. Refer to the Sun StorEdge Traffic Manager Software Installation and Configuration Guide for details. If you require automatic path switching for devices that are not supported, use Alternate Pathing software with DR model 2.0.

Sun Netra T1 Server

The Sun Netra trademark T1, a rackmounted thin server, can now be used as an SSP. For further information on the Netra T1, refer to the Netra t1 User and Administration Guide .

Changes to the SSP Software

SSP 3.5 offers:

Support for 466 MHz processors.
Performance improvements that enable the SSP to function continuously under adverse conditions, such as when CPU and memory utilization is high.

These improvements use real-time processes and enable the SSP to retry operations without creating further excessive load on the SSP workstation or server.
OpenSSP refers to the capability to run lightweight, third-party applications--other than the SSP software and the Solaris operating environment--on an SSP workstation or server.

However, be sure that your SSP meets the resource requirements for running third-party applications along with the SSP software, as explained in Chapter 7 .
There are two different models for dynamic reconfiguration. The DR model number indicates the type of DR functionality running on a Sun Enterprise 10000 domain:
- DR model 2.0 - Uses the dr_daemon to control DR operations on the domain. You can use Hostview, dr (1M) shell, or automated dynamic reconfiguration (ADR) commands on the SSP to perform DR operations.
  
  DR model 2.0 is the default for domains running the Solaris 8 10/01 (Update 6) operating environment. If you want to run DR model 3.0 (explained later) on a domain, refer to the instructions for switching DR models in the Sun Enterprise Dynamic Reconfiguration User Guide .
- DR model 3.0 - Uses the domain configuration server, dcs (1M) to control DR operations on the domain. You use the ADR commands on the SSP to perform DR operations. DR model 3.0 is consistent with DR functionality provided on other Sun server lines.
  
  DR model 3.0 also interfaces with the Reconfiguration Coordination Manager (RCM) feature, which enables you to coordinate DR operations with other applications running on the domain, such as database, clustering, and volume management software. Such applications can register to receive notification of DR operations and coordinate their software activities with DR. For details on RCM, refer to the System Administration Guide: Basic Administration in the Solaris 9 System Administrator Collection.

Note - DR model 3.0 is available only in the Solaris 8 10/01 and 02/02 or Solaris 9 operating environment. DR model 3.0 also requires that SSP 3.5 software be running on the SSP. Only DR 3.0 runs in a domain running the Solaris 9 operating environment. For instructions on installing or upgrading to the Solaris 8 or 9 operating environment, see Chapter 8 and Chapter 9.

For details on DR and how the DR models work, refer to the Sun Enterprise 10000 Dynamic Reconfiguration User Guide and the Sun Enterprise 10000 Dynamic Reconfiguration Reference Manual .

Changes to DR and SSP Commands and Behavior

This section summarizes the changes to commands and behavior for DR and SSP 3.5:

TABLE 1-1 explains the command changes related to DR.
TABLE 1-2 describes the changes in SSP commands.

TABLE 1-1 DR Command and Behavioral Changes
Command Name or Behavior	Applicable to DR Model	Description of Change
`addboard` (1M), `deleteboard` (1M), and `moveboard` (1M)	2.0 and 3.0	These automated dynamic reconfiguration (ADR) commands can be used to perform DR operations on domains running either DR model 2.0 or 3.0. Note the following changes in command behavior: `addboard` (1M) - If an error occurs in an `addboard` operation, an error message is written to standard error and the board is left in its failed state. You can either continue the operation by rerunning the `addboard` command or undo the operation by running the `deleteboard` (1M) command. `deleteboard` (1M) - If an error occurs in a `deleteboard` operation, an error message is written to standard error and the board is left in its failed state. You can either continue the operation by rerunning the `deleteboard` command or undo the operation by running the `addboard` (1M) command. `moveboard` (1M) - If an error occurs in a `moveboard` operation, an error message is written to standard error. If the specified board is detached and powered on, `moveboard` tries to attach the board to the specified domain. If the specified board is already in the specified domain, `moveboard` returns an exit status 0 (zero), indicating the operation was successfully completed. Each of these commands also offers the `-f` option, which attempts to force an operation by overriding certain software constraints. Note that these commands have additional exit status codes related to DR model 3.0. These commands, formerly explained in the Sun Enterprise 10000 SSP User Guide , are now described in the Sun Enterprise 10000 Dynamic Reconfiguration User Guide and the Sun Enterprise 10000 Dynamic Reconfiguration Reference Manual . For further details, see the corresponding man pages.
`domain_status` (1M)	2.0 and 3.0	The `domain_status` command now offers the `-m` option, which lists the DR model, either 2.0 or 3.0, used for each domain. For details, see the `domain_status` (1M) man page.
`rcfgadm` (1M)	3.0	`rcfgadm` (1M) enables you to remotely check DR model 3.0 board states from the SSP. For details, refer to the `rcfgadm` (1M) man page, the Sun Enterprise 10000 Dynamic Reconfiguration Reference Manual and the Sun Enterprise 10000 Dynamic Reconfiguration User Guide .
`showdevices` (1M)	3.0	`showdevices` (1M) lists system board devices and resource usage information. This command, while it can be used to obtain general device and resource information, is used primarily to obtain information on DR operations performed on DR model 3.0 domains. For details, refer to the `showdevices` (1M) man page or the Sun Enterprise 10000 SSP Reference Manual .
`showusage` (1M)	2.0	The `showusage` (1M) command is not supported on DR model 3.0 domains. The command has an exit status code that indicates the board belongs to a DR model 3.0 domain. For details, see the `showusage` (1M) man page or the Sun Enterprise 10000 SSP Reference Manual .

TABLE 1-2 SSP Command and Behavior Changes
Command	Description of Change
`bringup` (1M)	Any arguments specified with the `boot_args` option of the `bringup` (1M) command, must be enclosed by quotation marks. For details, see the `bringup` (1M) man page.
`setdatasync` (1M)	The file propagation status includes another category, `Active Archive` , which indicates that a data synchronization backup is in progress.
`setfailover` (1M)	The `setfailover` (1M) command checks the data synchronization state before forcing a failover and does not force the failover if: Data synchronization backup (active archive) is being performed File is being propagated One or more files exist in the data synchronization queue
`ssp_downgrade(` 1M)	Use the `ssp_downgrade` (1M) command to downgrade from SSP 3.5 to SSP 3.2. It is not necessary to run this command if you are downgrading from SSP 3.5 to SSP 3.4 or SSP 3.3.
`sys_id` (1M)	The ranges for the `-k` `key` , `-s` `serial no` , and `-h` `host` options of the `sys_id` (1M) command have been expanded.

Where to Find DR Information

The following table describes where you can find additional DR information.

TABLE 1-3 Where to Find DR Information
For information on	Refer to the
How DR models 2.0 and 3.0 work	Sun Enterprise 10000 Dynamic Reconfiguration User Guide
Switching the DR model for a domain	Sun Enterprise 10000 Dynamic Reconfiguration User Guide
Configuring DR Device prerequisites Overview of DR configuration tasks Prerequisites for DR detach operations Information on configuration changes during DR operations and how to control them How DR model 2.0 and Alternate Pathing software interact	Sun Enterprise 10000 DR Configuration Guide
Automated dynamic reconfiguration (ADR) commands: `addboard` (1M) `deleteboard` (1M) `moveboard` (1M) `showusage` (1M) `showdevices` (1M)	Sun Enterprise 10000 Dynamic Reconfiguration User Guide, command descriptions in the Sun Enterprise 10000 Dynamic Reconfiguration Reference Manual , or the corresponding man pages
Obtaining board state information on the SSP for DR model 3.0 domains	`rcfgadm` (1M) man page, command description in the Sun Enterprise 10000 Dynamic Reconfiguration Reference Manual , or the Sun Enterprise 10000 Dynamic Reconfiguration User Guide
Obtaining board state information on the domain (for DR model 3.0 domains)	`cfgadm` (1M) man page or the command description in the Solaris 8 or 9 Reference Manual
DR commands in the Solaris operating environment: `dcs(1M)` - DR model 3.0 daemon `dr_daemon(` 1M) - DR model 2.0 daemon `dr` (7D) - DR model 2.0 driver `ngdr` (7D) - DR model 3.0 driver `ngdrmach` (7D) - DR model 3.0 driver	Corresponding man pages or command descriptions in the Solaris 8 or 9 Reference Manual
DR error messages	Sun Enterprise 10000 Dynamic Reconfiguration User Guide , Appendix A.

Known SSP Bugs

This section describes software bugs for SSP 3.5 software that you may notice or that can affect the software operation. Also check the Solaris Release Notes Supplement for Sun Hardware for your Solaris operating environment for additional release notes pertaining to the Sun Enterprise 10000.

snmpd Memory Leaks During ADR Operations (4486454)

Memory leaks related to the snmpd daemon can occur after multiple addboard (1M), deleteboard (1M), or moveboard (1M) operations.

Workaround: If SSP performance is affected by this memory leak, stop and start SSP daemons by doing the following:

As superuser on the main SSP, type:

ssp# /etc/init.d/ssp stop

ssp# /etc/init.d/ssp start

machine_server Memory Leaks During Multiple hpost Operations (4493987)

Memory leaks related to the machine_server daemon can occur after multiple addboard (1M), deleteboard (1M), or moveboard (1M) operations.

Workaround: If SSP performance is affected by this memory leak, stop and start SSP daemons by doing the following:

As superuser on the main SSP, type:

ssp# /etc/init.d/ssp stop

ssp# /etc/init.d/ssp start

Known DR Bugs

This section describes software bugs for this version of DR software that you may notice or that can affect the software operation.

DDI_RESUME Failure Results in Failure of Subsequent DR Operations (4353719)

If a device fails to resume after a DR detach operation, the resume failure can prevent subsequent DR operations.

Workaround: None.

deleteboard(1M) Operation on a Board with no Memory Causes a Domain Running the Solaris 7 or Solaris 8 Operating Environment to Crash (4468171)

If you perform a

deleteboard (1M) operation
moveboard (1M) operation
Detach operation using dr shell

on a board that contains no memory, and the domains involved are running the Solaris 7 or Solaris 8 operating environment, the affected domains may crash or hang.

Workaround: Perform DR operations only on boards with memory.

Problem with oprom_checknodeid() during DR (4474330)

A panic can occur under certain circumstances when the /dev/openprom interface accesses the PROM device tree after a DR disconnect. The openprom driver caches node information that can become invalid after a DR disconnect. As a result, OBP can be passed a bad node address.

Workaround: To minimize the possibility of encountering this situation, discontinue using applications, such as prtconf , that use the /dev/openprom interface during or immediately before or after a DR disconnect operation. Note that picld (1M) uses the /dev/openprom driver.

showdevices(1M) Memory Information Displayed by showdevices Might Be Incorrect (4495747)

Domains running both DR 3.0 and Solaris 8 cause the showdevices (1M) command to display 0x0 as the base memory address.

Workaround: Obtain base memory address information by using the rcfgadm (1M) command.

Bugs Fixed Since SSP 3.4

The following bugs have been fixed in SSP 3.5 since SSP 3.4:

4365206 - Solaris 8 6/00, SSP 3.3 fails to install Ultra-Enterprise-10000.snmpd.cnf file

4365471 - ssp_restore overwrites newer files with older ones.

4365492 - Stopping the SSP during a cb_reset caused all domains to reset

4377087 - SSP failover in SSP 3.4 does not work correctly under certain circumstances

4378472 - SSP reports power supply 5 is missing even through having 4 power supplies is a legal configuration

4379020 - Domain reboot hangs if a new entry has been made in the blacklist file to exclude a component

4379395 - Unable to bring up a crashed domain after a control board failover in
SSP 3.4

4380350 - Hostview does not display the correct processor temperature for the appropriate system board

4383758 - config_float_ip ups the floating IP for the main SSP incorrectly

4402889 - Domain bringup does not verify that the boot processor returned from HPOST is in the domain

4405737 - Mechanical trouble with a fan tray can cause fan trays to power off

4411156 - " cpu tick sync failed in 256 attempt! " displayed during domain bringup on SSP 3.1.1

4415072 - " Bogus clk_mode " errors occur during phase proc1 for arbstop/recordstop/timeout

4416597 - Change SENSORMIN for system board temperature monitoring to reflect the onboard power supply redundancy

4417275 - procestemp reports "-0.00" temperature values

4423777 - Forced failover by "init 0" will often cause unexpected CB failover

4425460 - Exploitable static data buffer overflow with SSP snmpd

4428946 - Hostview force button is not enabled when xntpd is running

4451899 - All domains arbstopped after magic_cookie and libscan error reported to messages

4463339 - Failover does not reactivate a previously disabled spare properly

4469366 - Buffer overflow vulnerability in cb_reset

Requests for Enhancement Completed in SSP 3.5

The following Requests for Enhancement (RFE) have been implemented in SSP 3.5:

4268088 - Modify POST/SSP to reflect processor speed of 466 MHz

4351410 - Run SSP daemons as RT (real-time) process class

4360376 - autoconfig changes required to support new ecache SRAMs

4362960 - POST changes required to support new ecache SRAM

4366498 - Create new hpost ecache tests for new ecache SRAM

4368475 - Change check_host to log signatures when unexpected signatures are encountered

4376375 - Change libsysid to support new hostid ranges

4390086 - Need new client interfaces for Sun Enterprise 10000 DR model 3.0 support and other cleanups

4394348 - Disable setting of unused system reset bit in the control board (for reliability)

4396975 - Change SSP configuration scripts to automatically upgrade to PROM version 3.47

4402880 - ADR sample script should use showdevices (1M) command instead of showusage (1M) command to check result

4418571 - SSP needs to further check DR model 3.0/model 2.0 driver upon determining domain DR models

4419070 - Send a BBSRAM close net command upon netcon_server initialization

4419614 - Change default OS version to 5.8 in Hostview Create Domain window

4419788 - MSRAM processor property for Sun Enterprise 10000 system

4422147 - POST logs, POST dumps, arbstop and recordstop file names need a year component

4425374 - Enhance "unexpected foreign PIO" and "MC Timeout" messaging

4425380 - Remove dependency of ssp_startup on the presence of libnvpair

4458358 - POST ecache TAG test enhancement needed

4458552 - ssp_config (1M) should inform users to use lowest processor speed when processors are mixed

4470431 - sys_clock man page not clear for -s option

Patches

This section contains any required patches for SSP 3.5 software. Also check the Web site

http://sunsolve.Sun.com

on a regular basis for any SSP patches that you might need for SSP 3.5 and also any recommended patch clusters for the version of the Solaris operating environment that you are running.

If you need to install SSP software patches, be sure to install the patches on both the main and spare SSP and review To Prepare for SSP 3.5 Upgrade or SSP Patch Installation .

The following table lists the patches currently applicable to SSP 3.5:

TABLE 1-4 Required Patches for SSP 3.5
Patch Number	Install on
Solaris KU patches 106541-17 and 111422-01	SSPs running the Solaris 7 operating environment.
Solaris KU patch 110945-02	SSPs running the Solaris 8 6/00, 10/00, 1/01, or 4/01 operating environment (Solaris updates 1, 2, 3, and 4)

Reference Manual (man pages) Errata

These additions or errors apply to the following man pages:

autoconfig (1M) - Add to this command to "Perform this command when a memory mezzanine, I/O mezzanine, processor module, or system board is replaced with another component that has a different part number. Then reboot the SSP".
deleteboard (1M) - Add to this command, exit code 16 "bringup -C requested process termination" .
moveboard (1M) - Exit code 10 for this command should read "Failed in attempting to complete the detach" .

Add to this command, exit code 16 "bringup -C requested process termination" .
netcon (1M) - List of -e escape characters should be: @ # ^ & * = . ?