PCI Error Handling Product Note 3rd Edition

and the same (or later) release version number, then repeat the Post Replace operation
described in Step 2.
4. If the Post Replace operation succeeds and the I/O card/slot recovers from the error, the
software state of the components will be marked CLAIMED in the ioscan(1M) output. If
you continue to experience errors on this slot, there is a high probability that the I/O card is bad. HP
recommends replacing the I/O card with an I/O card that has the same HP Manufacturing
Part Number and the same (or later) release version number, then repeat the Post Replace
operation described in Step 2.
IMPORTANT: If you use Serviceguard, HP recommends the PCI Error Handling feature only
be enabled if your storage devices are configured with multiple paths and are protected by high
availability storage software such as PVLink, SecurePath, or MirrorDisk/UX. If PCI Error Handling
is enabled, but your storage devices are configured with only a single path, a system reboot may
be necessary to recover from a PCI error.
NOTE: With the PCI Error Handling solution installed, there is still a remote possibility that
an MCA or HPMC could occur during a PCI OLA operation (online addition of an I/O card). At
the beginning of a PCI OLA operation, there is a brief time during which the PCI Error Handling
infrastructure determines if the driver associated with the card is PCI Error Handling capable.
Any PCI error that occurs during this brief window of exposure can cause an MCA or HPMC.
This exposure only exists during PCI OLA operations. This exposure does not exist during PCI
OLR operations (online replacement of an I/O card), or during ordinary I/O card operations.
The following example shows how the PCI Error Handling feature is used to handle a PCI error
involving the iether driver:
NOTE: The PCI Error Handling procedure detailed in this example may vary slightly from
what you will experience, depending on the platform and IO card driver.
A. A PCI error occurs and error messages are displayed on the console:
-------------------100BT/Gigabit Ethernet LAN/9000 Networking---------------@#%Thu Jan 24 MST 2008
21:50:49.540624 DISASTER Subsys:IETHER Loc:00000<1002> 1000Base-T in path 6/0/0/1/0 Was
moved to DEAD state due to a PCI
error.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-------------------100BT/Gigabit
Ethernet LAN/9000 Networking---------------@#%Thu Jan 24 MST 2008 21:50:49.565469 DISASTER
Subsys:IETHER Loc:00000<1004> 1000Base-T in path 6/0/0/1/0 Is being suspended due to a PCI
error.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-------------------100BT/Gigabit
Ethernet LAN/9000 Networking---------------@#%Thu Jan 24 MST 2008 21:50:49.585899 DISASTER
Subsys:IETHER Loc:00000<1004> 1000Base-T in path 6/0/0/1/1 Is being suspended due to a PCI
error.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
B. Execute the olrad -q command to confirm the card is in the suspended state:
How to Online Recover from a PCI Error 13