FUJITSU United Kingdom

Worldwide

Site Map

skip the main navigation

> Support & Downloads > Infrastructure > VME > VME IONS

VME Support

Identifier	S194	Issue	2
Title	Enhancements to MAMPHY to Provide Additional Information.
Priority	B
Date	2007/08/06

Enhancements to MAMPHY to Provide Additional Information.

A set of repairs has been issued at Open VME Version 6 to primarily highlight rare but important conditions to the user through software incident reporting. In addition the repairs provide extra diagnostic information for use by Fujitsu Services support staff in the event that a problem requires investigation.

At Open VME6, this set of repairs can be applied to your loadset by

1) retrieving the latest Open VME 6 repair envelope, and bringing that loadset to the latest RRL, which will include having taken note of the Special Instructions for the Edit repairs MM/K/1029 and MM/K/1030 before applying them.

2) applying optional repairs MM/K/1024 and MM/K/1025 if required
(please see section 1(d) for further information).

At Open VME7 all repairs except for MM/K/1024 and MM/K/1025 are pre-applied.

Repairs that must be applied:
MM/K/1029.1 MM/K/1030.1	To introduce the new software incident message texts (and a prompt) as required by the following repair set.
VU/K/0522.1	These repairs will be automatically applied using the above instructions.
MM/K/0925.1
MM/K/0926.1

Enhancement Repairs:
MM/K/1007.2	For software incident reporting on FLF transfer failures.
MM/K/1031.1	For software incident reporting on FLF transfer failures.
MM/K/1008.2	For software incident reporting on PVD transfer failures.
MM/K/1009.2	For software incident reporting on Locking failures due to too many block locks on a file.
MM/K/1010.1	For additional tracing during especially the VME loading process for problem investigations by Fujitsu Services.
MM/K/1011.1
MM/K/1012.1
MM/K/1013.1
MM/K/1014.2	For additional tracing concerning possibilities of transfer overload for problem investigations by Fujitsu Services.
MM/K/1015.2	For software incident reporting on PVD anomalies.
MM/K/1016.2
MM/K/1017.2
MM/K/1018.2
MM/K/1019.2
MM/K/1020.2
MM/K/1021.2
MM/K/1022.2
MM/K/1023.1	For anomalies in increasing the size of the lock table.
MM/K/1026.1	For anomalies in increasing the size of the lock table.
MM/K/1027.1	To provide additional information in DMMT output for problem investigations by Fujitsu Services.
MM/K/1028.1
MM/K/1024.2	Optional repairs providing software incidents for instances of 'lock table full'. See section 1(d) for details.
MM/K/1025.2

For background information, the following sections give details of the new message texts and any actions to be taken.

Should you require further assistance then please contact your local Fujitsu Services support staff.

1. SOFTWARE INCIDENT REPORTING:

New messages are reported to highlight existing and potentially critical situations to enable their early resolution. These system problems relate to:

the PVD of a disc. If a PVD transfer fails or any anomalies are encountered within the PVD then partition actions requiring write access to the PVD are inhibited to maintain its integrity.
the FLF of a partition. If an FLF transfer failure occurs then jobs requiring that partition may fail.
limit conditions within the locking mechanism. If a lock table is full then jobs may fail but the underlying cause may be due to a breach of the recommended locking actions.

a) To Introduce the New Software Incident Messages:

Message Text Repairs:

MM/K/1029.1,

MM/K/1030.1

Purpose:

These repairs introduce the new messages which are described in the sections below. The Special Instructions sections of these repairs identify the actions to be taken for their application.

b) For PVD Problems:

i) Repair:

MM/K/1008.2

Software incident message:

INCIDENT ON DISC yyyyyy, READ/UPDATE TRANSFER TO PVD WITH
FSI aaaa FROM FIXED BLOCK nnnnnnnnn

Reason:

This message is reported to highlight that a transfer failure has occurred on the PVD of the disc yyyyyy, which has the FSI of aaaa . The first fixed block in the failing transfer is reported as nnnnnnnnn, however it should be noted that the transfer may have been a multiblock transfer. The software incident failure code will provide the MAMPHY underlying result code.

Action:

Currently Hardware Error Logs and MSRs are raised for transfer failures, this reporting remains unchanged. The additional software incident is to specifically alert the system manager and operations staff to potentially critical issues. BN5114010 - OpenVME: Using and Managing Disc Systems, Section E: ‘Recovering from Disc Problems' may be consulted for additional information. However it is NOT advisable to carry out any procedure on the PVD without fully understanding the nature and cause of any underlying problem. You are advised to contact your local Fujitsu Services Support Centre for assistance.

ii) Repairs:	MM/K/1015.2,	MM/K/1016.2,	MM/K/1017.2,	MM/K/1018.2,
	MM/K/1019.2,	MM/K/1020.2,	MM/K/1021.2,	MM/K/1022.2

Software incident message:

PARTITION ACTIONS INHIBITED ON DISC yyyyyy, PROBLEM
ENCOUNTERED WITH MASTER/SLAVE/MASTER AND SLAVE PVD, TRACE
POINT hh

Reason:

If this failure is reported then it may be additional to information registered in the hardware error log , an MSR and possibly a software incident for a transfer reading or updating the PVD. However it may also be reported without any of these if the PVD cell that has been read has an anomaly. The software incident failure code will report the MAMPHY result code.

The message identifies which copy/copies of the PVD it suspects is/are corrupt and that the failure was to the PVD of the disc yyyyyy. In addition a trace point identifier hh is reported for possible use by Fujitsu Services support staff.

Action:

It is NOT advisable to carry out any procedure on the PVD without fully understanding the nature and cause of any underlying problem. You are advised to contact your local Fujitsu Services Support Centre for assistance.

c) For FLF Transfer Failures:

i) Repairs:

MM/K/1007.2 ,

MM/K/1031.1

Software incident message:

INCIDENT ON PARTITION xxxxxx, READ/UPDATE TRANSFER TO yyyyyy BLOCK/S FROM FLF CELL nnnnnn FAILED

with the additional text on the 7th incident per partition establishment:

NO MORE SUCH INCIDENTS WILL BE LOGGED FOR THIS PARTITION UNTIL IT IS RE-ESTABLISHED.

Reason:

The message is reported to highlight that a transfer failure has occurred on an FLF of the partition xxxxxx, that it was either a single block request or a multiple block request to yyyyyy blocks and it provides the start FLF cell number nnnnnn. The software incident failure code will provide the MAMPHY underlying result code.

Should there be more than one occurrence of an FLF transfer failure whilst a partition is established for use then only the first such incident will appear in the Operator's software incident picture and up to 7 occurrences will be registered in the system journal per establishment. Should the number of incidents exceed 7 then reporting will be suspended and will not resume until either the partition is re-established or the time reaches midnight. The count of incidents on established partitions is reset to zero at midnight providing that this is not prevented by their usage.

Multiple copies for the same cell may appear.

Action:

Currently Hardware Error Logs and MSRs are raised for transfer failures, this reporting remains unchanged. The additional software incident is to specifically alert the system manager and operations staff to potentially critical issues. For guidance on resolving any such matters, BN5114010 - OpenVME: Using and Managing Disc Systems, Section E, ‘Recovering from Disc Problems' should be consulted. Should additional assistance be necessary then customers are advised to contact their local Fujitsu Services support staff.

d) For Locking Problems:

In addition to highlighting a limit condition being encountered on the lock table, these software incidents will enable the early dumping of that table if this is required. This can be achieved using

DMMT(LOCK, FIL=pre-created_empty_disc_file)

Where a particular file is identified in the message and the underlying cause of the software incident is not understood then additional information may still be available for Fujitsu Services support staff to investigate if the partition on which the file resides is dumped immediately using

DMMT(VOLUME, partition_name, FIL=pre-created_empty_disc_file)

Two of the repairs raised for locking problems are optional to ensure that any site which expects such messages and has a process in place to cater for them need not be subjected to this additional information.

i) Optional repair:

MM/K/1024.2

Software incident message:

INCIDENT ON MM_LOCK_TABLE, NO MORE FILE SECTIONS CAN
CURRENTLY BE LOCKED

Reason:

This message is reported when an attempt is made to attach a file section for locking and the current number of file sections locked has already reached the maximum limit of 3069 (MMP 5 as specified in section 2.4.6 of BN519098 OpenVME: System Limits and Initialisation Options).

A further file section can only be attached for locking once one of the currently locked sections is detached.

The failure code for the software incident is MM_MAMPHY_LOCK_TABLE_FULL.

ii) Optional repair:

MM/K/1025.2

Software incident message:

INCIDENT ON MM_LOCK_TABLE, NO LOCK RECORDS CURRENTLY
AVAILABLE

Reason:

This message is reported when a block lock requested for a file cannot be satisfied.

The limit MMO 3 (specified in section 4.1.3 of BN519098 OpenVME: System Limits and Initialisation Options) is 14 locks per section for efficiency as a lock record will be allocated initially from its own area which may be controlled by up to 3 files sections. When this area is exhausted further locks are allocated from areas with free file section locks.

A further block lock for this section will only be possible once other locks are released and possibly another currently locked section is detached.

The failure code for the software incident is MM_TF_LOCK_TABLE.

iii) Repair:

MM/K/1009.2

Software incident message:

INCIDENT ON MM_LOCK_TABLE, LOCKS ON SECTION EXCEEDED, FILE UNKNOWN/' hex representation of the reverse link of the file'

Reason:

This message is reported when a block lock on an attached section is requested and the current number of the owned plus the waiting locks is already at the maximum permitted number of block locks per section. This limit is 4095 and will in future be referenced as a locking limit of MMP 14.

Wherever the name of the file is available it will be reported as the hex representation of the file section name contained within the first 29 bytes of the reverse link. In the case of a multi section file this may possibly be for the first section. The hex characters may be decoded using the EBCDIC character set but bear in mind that non alphanumeric characters may be compressed/omitted such that the top 2 bits of the following alphanumeric character will be zero.

If the name of the file is not available then ‘UNKNOWN' will be reported.

The failure code for the software incident is STD_TABLE_FULL.

Should further assistance be necessary then please contact your local Fujitsu Services support staff for advice.

2. ADDITIONAL MAMPHY TRACE INFORMATION

These repairs provide additional trace information with regard to MAMPHY during the VME loading process and MAMPHY's transfer overload processes. This information may be requested by Fujitsu Services support staff to assist in any problem resolution. It may be obtained using the command DUMP_MAMPHY_TABLES (i.e. DMMT) as defined below.

i) Repairs:

MM/K/1010.1,

MM/K/1011.1,

MM/K/1012.1,

MM/K/1013.1

Purpose:

Should there be any issues with the loading process, for instance the system load has taken an unexpectedly long period of time then, in addition to any other diagnostic information that may be required, the user should also submit the output file from a call of

DMMT(MMUM_TRACE_TABLE,FIL=pre-created_empty_disc_file)

to Fujitsu Services support staff for analysis.

Additional Software Incident:

USING MINIMUM SIZE FOR MMUM_TRACE_TABLE

This set of repairs has the potential to report the above Software Incident in the System Journal and the Software Incident Operator's Picture File if the system failed to provide for an extended length MMUM trace table to cater for the additional tracing. In these circumstances the tracing will default to the minimum sized table.

Action:

This software incident is purely for information only, no action is required by the operations staff.

ii) Repair:

MM/K/1014.2

Purpose:

This repair provides additional tracing information logging any transfer overload or event bottlenecks occurring within a particular VM. It will be available in dumps from calls of DMMT such as

DMMT(	TABLE = MMDI, @ or ALL_DATA @ FILE = pre-created_empty_disc_file, VIR = vm_number_of_the_vm_in_question (or 1 where no particular VM is identified)
)

This information would normally be requested by Fujitsu Services support staff when investigating an Open VME system problem reported by the user.

iii) Repairs:

MM/K/1027.1,

MM/K/1028.1

Purpose:

These repairs provide for additional information to be available in the dumps for calls of DMMT, in particular with respect to:

DMMT(	TABLE = MMDI, @ or MMDV or ALL_DATA @ FILE = pre-created_empty_disc_file, VIR = vm_number_of_the_vm_in_question (or 1 where no particular VM is identified)
)

This information would normally be requested by Fujitsu Services support staff when investigating an OpenVME system problem reported by the user.

3. ADDITIONAL MAMPHY REPAIRS

i) Repairs:

MM/K/1023.1,

MM/K/1026.1

Purpose:

These repairs correct anomalies in the current reporting that may potentially arise when dynamically increasing the required size of a lock table. They do not introduce any new messages

4. NEW SYSTEM PROMPT:

Once the message text edits have been applied together with any of the MAMPHY enhancement repairs in the range MM/K/1007 to MM/K/1008 and MM/K/1015 to MM/K/1022 inclusive, then the following prompt may be reported to the operator:

MM S/W INCIDENT, SEE SI OPF

Operator Action
Press send to acknowledge receipt of the prompt and clear the prompt from the screen.

Further Information:

The SI Operator Picture (OPF) and the System Journal will provide information clarifying the problem as described in section 1 (Software Incident Reporting).

Should further assistance be necessary then please contact your local Fujitsu Services support centre for advice.

Reason:

This prompt will be output when reporting the new software incidents for:

Any PVD problem.
For the first occurrence of an FLF transfer problem whilst the partition is established. The count of incidents on established partitions is reset to zero at midnight providing that this is not prevented by their usage and thus this prompt has the potential for being reported on a daily basis for partitions that remain active for long periods of time.

5. FURTHER RELATED DIAGNOSTIC FACILITIES

Please see ION S200 for information concerning other related diagnostic facilities applicable at OVME7GR.

click here to return to index

Comments are welcome and may be sent to:
VME Customer Care Unit Fujitsu Services Central Park Northampton Road Manchester M40 5BP England	E-mail:VME.support@uk.fujitsu.com Telephone: +44 (0)870 325 3354 Fax: +44 (0)870 325 3526
Find out more now To find out more about VME contact us by using the feedback form or call Fujitsu Services on +44 (0)870 242 7998, Fax +44 (0)870 242 4445. Copyright © Fujitsu Services Limited 2008 All rights reserved. Copyright in this document remains vested in Fujitsu Services and no copies may be made of it or any part of it except for the purpose of evaluation in confidence.

About

Privacy

back to the main navigation