IBM eServer Cluster 1600 Managed by PSSP 3.5: What's New

136 IBM ^ Cluster 1600 Managed by PSSP 3.5: What’s New

3. After the migration, we re-established the switch communication and ran

verification tests. We did not have any problems with these steps.

4. Install new versions of VSD and GPFS and run tests on the file systems. For

VSD, we installed the bos.clvm.enh fileset as a prerequisite.

6.3.2 Migrating PSSP 3.1.1 and AIX 4.3.3 to PSSP 3.5 and AIX 5.1

For this scenario, we chose a system with four 112 MHz SMP high nodes

connected by an SP Switch. AIX 4.3.3 with Maintenance Level 10 and PSSP

3.1.1 are installed on the CWS and on the nodes. We have the first node

configured as an NFS server with a file system that is mounted on the rest of the

nodes.

There is no direct migration path available from the software level we have on this

cluster. Therefore, we decided to move to PSSP 3.4 first on both the CWS and

the nodes, and then to AIX 5L Version 5.1 ML3 and PSSP 3.5. For the first part,

we followed the

PSSP for AIX: Installation and Migration Guide

, GA22-7347 for

PSSP 3.4 and then for PSSP 3.5.

We completed the following steps:

1. Migrate the CWS to PSSP 3.4:

– Before migration, check for non-ASCII data in the SDR by running

SDRScan. Verify system configuration and

connectivity

.

– After migration, we found that

spmon and spsvrmgr do not give back

information about the nodes. The following examples contain the related

error messages we found. General messages from SPdaemon.log are

shown in Example 6-4 on page 137.

Attention: PSSP 3.1.1 is not supported by IBM. Direct migration from PSSP

3.1.1 to PSSP 3.5 is not supported. We developed this section to show that it

is possible to use the latest level of PSSP software with older SP nodes as

well.

Important: To avoid the following problem in “Step 22: Run SbR and

system monitor verification test” in Chapter 4 of the

PSSP for AIX:

Installation and Migration Guide

, GA22-7347, reset the Hardware Monitor

daemon by running the

hmreinit command. This is correctly documented

in step 20 of the

PSSP for AIX: Installation and Migration Guide

,

GA22-7347.

Chapter 6. Coexistence, migration, and integration 137

Example 6-4 Error message in /var/adm/SPlogs/SPdaemon.log

Oct 4 13:41:10 sp3cws hardmon[32052]:

LPP=PSSP,Fn=hm_tty.c,SID=1.21.4.15,L#=264, hardmon: 0026-850 Data length

mismatch in packet from tty /dev/tty0 (Frame 1): calculated = 15, received =

14.

Oct 4 13:42:09 sp3cws sphwlog[32106]:

LPP=PSSP,Fn=splogd.c,SID=1.16.7.3,L#=1537, 0026-107 Failure; Frame 1:0;

frPowerModCbad; Power module - DC power loss.

The SDR configuration messages are shown in Example 6-5.

Example 6-5 SDR_init log file: /var/adm/SPlogs/sdr/SDR_config.log

SDR_init: SDR_init was invoked at Fri Oct 4 15:36:04 EDT 2002 with flag values

of debug=0, log=1 and verbose=0.

SDR_init: 0016-082 An error has been encountered while internally executing the

command "/usr/lpp/ssp/bin/hmmon -Q -v type -r 1,5,9,13 2> /dev/null". The

return code from the command was 1. SDR_init is continuing.

SDR_init: 0016-082 An error has been encountered while internally executing the

command "/usr/lpp/ssp/bin/hmmon -Q -v type -r 1,5,9,13 2> /dev/null". The

return code from the command was 1. SDR_init is continuing.

SDR_init: 0016-705 Problem while attempting to read Hardmon data.

SDR_init: 0016-733 SDR_init completed unsuccessfully with a return code value

of 2.

Hardmon-related messages are shown in Example 6-6.

Example 6-6 Hardmon daemon log file /var/adm/SPlogs/spmon/hmlogfile.277

hardmon: 0026-801I Hardware Monitor Daemon started at Fri Oct 4 13:20:31 2002

hardmon: 0026-802I Server port number is 8435, poll rate is 5.000000 seconds

hardmon: 0026-805I 1 frames have been configured.

hardmon: 0026-803I Entered main processing loop 0000001c

hardmon: 0026-808I Received command to quit from SIGTERM at sp3cws/0.

hardmon: 0026-801I Hardware Monitor Daemon started at Fri Oct 4 13:27:39 2002

hardmon: 0026-802I Server port number is 8435, poll rate is 5.000000 seconds

hardmon: 0026-805I 1 frames have been configured.

hardmon: 0026-803I Entered main processing loop 00000038

hardmon: 0026-808I Received command to quit from SIGTERM at sp3cws/0.

hardmon: 0026-850 Data length mismatch in packet from tty /dev/tty0 (Frame 1):

calculated = 15, received = 14.

– After restarting the hardmon daemon with the

/usr/lpp/ssp/install/bin/hmreinit command on the CWS, everything

worked fine.