Interface Card Critical Resource Analysis(CRA) Whitepaper for HP-UX 11i v3

ManualsBrandsHP ManualsSoftwareHP-UX 11i v3 Virtual Server Operating Environment (VSE-OE) LTU

Critical Resource Analysis (CRA) White Paper

SUMMARY ......................................................................................................................................... 3

Introduction to Critical Resource Analysis................................................................................................ 4

CRA return values ............................................................................................................................ 4

Mass Storage CRA Scenarios................................................................................................................ 5

Boot Path Configuration Scenarios ..................................................................................................... 5

Scenario 1 .................................................................................................................................. 5

Scenario 2 .................................................................................................................................. 6

Scenario 3 .................................................................................................................................. 6

Scenario 4 .................................................................................................................................. 7

Swap and Crash Dump Device Configuration Scenarios....................................................................... 8

Scenario 5 .................................................................................................................................. 8

Scenario 6 .................................................................................................................................. 8

Scenario 7 .................................................................................................................................. 9

Service Guard Configuration Scenarios .............................................................................................. 9

Scenario 8 .................................................................................................................................. 9

Scenario 9 ................................................................................................................................10

Scenario 10 .............................................................................................................................. 11

Scenario 11 .............................................................................................................................. 11

Scenario 12 .............................................................................................................................. 12

I/O in Progress Scenarios ............................................................................................................... 12

Scenario 13 .............................................................................................................................. 12

Scenario 14 .............................................................................................................................. 12

File System Configuration Scenarios................................................................................................. 13

Scenario 15 .............................................................................................................................. 13

Scenario 16 .............................................................................................................................. 13

Scenario 17 .............................................................................................................................. 14

Scenario 18 .............................................................................................................................. 14

Scenario 19 .............................................................................................................................. 15

Scenario 20 .............................................................................................................................. 15

DLKM Driver unload scenarios......................................................................................................... 16

Scenario 21 .............................................................................................................................. 16

Scenario 22 .............................................................................................................................. 16

Scenario 23 .............................................................................................................................. 17

Summary of content (31 pages)

PAGE 1
Critical Resource Analysis (CRA) White Paper SUMMARY ......................................................................................................................................... 3 Introduction to Critical Resource Analysis................................................................................................ 4 CRA return values ............................................................................................................................ 4 Mass Storage CRA Scenarios......
PAGE 2
Cards in PCI Error and Suspended State Scenarios ............................................................................ 17 Scenario 24 .............................................................................................................................. 17 Scenario 25 .............................................................................................................................. 18 Scenario 26 .......................................................................................
PAGE 3
SUMMARY Critical Resource Analysis (CRA) provides usage analysis for I/O resources during destructive operations, such as online PCI card replacement and deletion, lun or lunpath removal, and during unload of an interface card driver. The complete CRA framework consists of multiple subsystem CRA modules, and one infrastructure CRA module. The infrastructure CRA module provides basic functionality and co-ordination between the subsystem CRA modules.
PAGE 4
Introduction to Critical Resource Analysis Starting with HP-UX 11i v3, given a set of I/O components, the CRA module performs usage analysis and reports any potential impacts for specific subsystems, such as Mass Storage, LAN, or Infiniband. The CRA module returns the highest critical usage reported. The CRA module also collates detailed analysis logs from all the subsystem CRA modules and returns the combined logs for the user under the /var/adm directory on the system.
PAGE 5
olrad command provides an option to override the flag and proceed with the removal of I/O components. For example, if an IP address is configured on a networking port, it is considered a DATA CRITICAL usage. CRA_SYSTEM_CRITICAL - Indicates critical resources are in use. If these resources are taken offline, the system could crash. Swap and boot configurations are examples. With system critical resources in use, the corresponding I/O component should not be allowed to be taken offline.
PAGE 6
# scsimgr disable –H Criticality reported by CRA: CRA_SYSTEM_CRITICAL Explanation and user action required: The olrad(1M) command with –C option reports the resource usage and its criticality. The above operations with scsimgr and olrad (1M) with –r or –d options (if it were to be allowed) will render the boot lunpath unavailable. This operation will result in system crash, hang, or any other indeterminate system state.
PAGE 7
Criticality reported by CRA: CRA_DATA_CRITICAL Explanation and user action required: The olrad(1M) command with –C option reports the resource usage and its criticality. The above operations with scsimgr and the olrad(1M) with –r or –d options will fail with CRA_DATA_CRITICAL error.
PAGE 8
However, running the above kconfig –H command will discard all the kernel configuration changes, not only the kernel configuration change caused by the above kcmodule driver unload operation, but also any other changes held for next boot of the system.
PAGE 9
The olrad(1M) command with –C option reports the resource usage and its criticality. The olrad(1M) with –r or –d options (if it were to be allowed) will render the swap device unavailable and this will result in system crash, hang, performance issues, or any other indeterminate state. Because of the above reasons, the requested olrad operations will not be allowed by CRA to proceed and the operation will fail with CRA_SYSTEM_CRITICAL error.
PAGE 10
Criticality reported by CRA: CRA_SYSTEM_CRITICAL Explanation and user action required: The olrad(1M) command with –C option reports the resource usage and its criticality. The above operations with scsimgr and olrad(1M) with –r or –d options (if it were to be allowed) will render the quorum disk unavailable if the requested operation is allowed to proceed. This process will result in the node being evicted from the cluster.
PAGE 11
stop the SG cluster service, a different quorum disk can be configured using an alternative healthy card and the required olrad operation can be performed. Scenario 10 System Configuration: HP-UX OS running in Service Guard (SG) cluster environment with common lock disk between the cluster nodes, and the cluster is configured with a common lock disk using path dependent device special file.
PAGE 12
Scenario 12 System Configuration: HP-UX OS running in Service Guard (SG) cluster environment with common lock disk between the cluster nodes, but the SG cluster is not currently running.
PAGE 13
Explanation and user action required: The above operation with the rmsf option will fail with a CRA_DATA_CRITICAL error. Before performing the above operation, it is recommended to make the user applications use the path independent device special file. File System Configuration Scenarios Scenario 15 System Configuration: A File system is mounted on a disk and the disk has a single lunpath through a card.
PAGE 14
Scenario 17 System Configuration: Disk has been configured as a mirror copy of a logical volume (could be either LVM or VxVM volume) with file system mounted on it. The mirror disk has only one lunpath, where the lunpath originates from a different I/O card as compared to the primary mirror copy.
PAGE 15
However, if there is any process that is keeping the mirror disk open, then the above operations will return the CRA_DATA_CRITICAL error. If the user chooses to proceed with the requested operation or overrides the error (that is, in case of CRA_DATA_CRITICAL error), it may result in system failure and/or loss of data, in the event of failure of the primary mirror copy.
PAGE 16
DLKM Driver unload scenarios Scenario 21 System Configuration: A logical volume (could be either LVM or VxVM volume) has been created with two or more disks (that is, PVs), each PV has a single lunpath through same or different cards and a file system is residing on that logical volume.
PAGE 17
Scenario 23 System Configuration: HP-UX OS running in Service Guard (SG) cluster environment with common lock disk between the cluster nodes and the cluster is configured with a common lock disk, which has multiple lunpath(s) through different cards, controlled by different I/O drivers.
PAGE 18
Scenario 25 System Configuration: HP-UX OS is running and an interface card in one of the PCI I/O slots is in PCI ERROR state. That is, ioscan(1M) displays the nodes below the slot in ERROR state.
PAGE 19
SAN Storage Configuration Scenario Scenario 28 System Configuration: A set of LUNs is configured on the storage device to be visible to a particular fiber channel world wide name (WWN), which belongs to a fiber channel HBA on a system in the SAN. Possible usages of these LUNs could be: • A file system is mounted on one of the LUNs. • An I/O is in progress to one of the LUNs. • This LUN is being used as a dump device.
PAGE 20
However, if the user wants to override the CRA criticality report, he can do so at his own risk by using the –f option of the olrad command. If any of these operations must be performed for system administration purposes, the system must be made accessible through alternative network interface card, which does not have IP address already configured. To do this, first verify that the alternative NIC has a physical connection to the network. This can be verified using one of these commands: nwmgr, linkloop.
PAGE 21
configuration, re-assign the IP address to an alternative Network interface card (NIC) and then proceed with the olrad deletion. Scenario 3 System Configuration: HP-UX System is running, which has a multi-port Network interface card (NIC), with one or more NIC ports configured with IP addresses and connected to network. User Scenario: Replace the card with another card of same model. Or Delete and add a different type of card which can be claimed by the same or different driver.
PAGE 22
Because of the above reasons, the requested unload operation will not be allowed by CRA to proceed and the operation will fail with CRA_DATA_CRITICAL error. If this operation must be performed for system administration purposes, the user must re-assign all the IP addresses configured on the ports of the NIC, for which the driver is about to be unloaded to alternative NIC card(s) controlled by different NIC driver than the one about to be unloaded.
PAGE 23
Operation Performed: # olrad –C | -r | -d Criticality reported by CRA: CRA_DATA_CRITICAL Explanation and user action required: The olrad(1M) command with –C option reports the resource usage and its criticality. The above operations with olrad(1M) with –r or –d option (if it were to be allowed) will render the system inaccessible to the network.
PAGE 24
To delete and add a different type of card which can be claimed by the same or different driver. Operation Performed: # olrad –C | -r | -d Criticality reported by CRA: CRA_WARNING Explanation and user action required: The olrad(1M) command with –C option reports the resource usage and its criticality.
PAGE 25
User Scenario: Unload the interface driver of the LAN card, whose lan ppa is part of the APA configuration. Operation Performed: # kcmodule =unused Criticality reported by CRA: CRA_WARNING Explanation and user action required: The above operation with kcmodule(1M) will unload the NIC driver of LAN ports configured in the APA group, which will result in deletion of these LAN ports from the APA group.
PAGE 26
Explanation and user action required: The above nwmgr(1M) operations with –-delete option (if it were to be allowed) will delete the NIC ports in the link aggregate, which may result in network applications running on the system hanging, inaccessible to the network client applications, leading to application downtime or any other indeterminate system state.
PAGE 27
Criticality reported by CRA: CRA_DATA_CRITICAL Explanation and user action required: The above nwmgr(1M) operation with –-delete option (if it were to be allowed) will delete the NIC port in the link aggregate, which may result in the reduction of the available network I/O bandwidth of the APA group and may impact the LAN failover capability. Because of the above reasons, the requested nwmgr operation will not be allowed by CRA to proceed and the operation will fail with CRA_DATA_CRITICAL error.
PAGE 28
Operation Performed: # olrad –C | -d Criticality reported by CRA: CRA_SYSTEM_CRITICAL Explanation and user action required: The olrad(1M) command with –C option reports the resource usage and its criticality. The above operation with olrad(1M) with –d will remove either the SG heart-beat or the stand-by LAN connection, leading to loss of highly-available cluster functionality.
PAGE 29
Dynamic nPartitions Scenarios I/O Chassis Deletion Scenarios Scenario 1 System Configuration: HP-UX system is running on a cell-based system with multiple cell boards, and one or more of the cell board(s) is designated as floating cell, with I/O chassis connected to one of the floating cells. User Scenario: To replace or perform hardware maintenance online on a floating cell connected to an I/O chassis, which has active I/O usage.
PAGE 30
is necessary to use the ifconfig(1M) command with the unplumb option to tear down the networking configuration. NOTE: The Dynamic nPartition configuration guideline suggests that the I/O chassis associated with a floater cell not be configured to reach any critical resources, without additional redundant connections to the network or to the storage resources through a different I/O chassis.
PAGE 31
For more information Interface Card OL* Support Guide - HP-UX 11i v3 http://docs.hp.com/en/5992-1723/5992-1723.pdf Call to action HP-UX 11i v3 for HP Integrity and HP 9000 servers www.hp.com/go/hpux11iv3 © 2008 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services.