Service Guard Resource Monitoring

[Ron Cornwell]

I have (2) A500's running 11.11, with M/C SG A.11.14 connected to a VA7400 via a single Tachyon FC Adapter and dual silkworm fiber switches. I am trying to figure out how to have serviceguard monitor the disk/connection past the controller card. I believe I have it configured so that if the card is anything other than up serviceguard will fail to the other system. This LUNs show up as disk in an ioscan, but the do not show up as disk resources in Resource Management in SAM. Here is what I am currently monitoring:

/adapters/status/FC_adapter/0_4_0_0.1
/adapters/status/TL_adapter/0_4_0_0.1.16.0.0

I have tested pulling the fiber cable out of the FC adapter (leaving the GBIC install) to simulate a fiber cut. When I do a cmviewcl -v -p package1, it shows both those resources as still up leading me to believe I am not monitoring everything that I should.

Thanks in Advance.
Ron Cornwell

[Schelstraete Bart_1]

Hello,

Can you give us your cluster configuration? (the way the you configured MC/Serviceguard to monitor those resources)

Can you also give me the out put of resls:
#> resls /adapters/status/FC_adapter/0_4_0_0.1
#> resls /adapters/status/TL_adapter/0_4_0_0.1.16.0.0


rgrds,

Bart

[Ron Cornwell]

NAME: /adapters/status/FC_adapter/0_4_0_0.1
DESCRIPTION:
UP: A state of "UP" means the hardware is available for use.

DOWN: A state of "DOWN" means the hardware is unavailable for use.
The /opt/resmon/bin/set_fixed command must be used with the resource name once the hardware is fixed to reset the hardware to the "UP" state.

UNKNOWN:
A state of "UNKNOWN" means it is unknown whether the hardware is unavailable for use or not. The hardware should be considered DOWN.

TYPE: /adapters/status/FC_adapter/0_4_0_0.1 is a Resource Instance whose values are enumerated values.

There are 3 valid states reported for /adapters/status/FC_adapter/0_4_0_0.1:

State Value State Name
----------- ----------
0 UP
1 DOWN
2 UNKNOWN

There is 1 active monitor request reported for this resource.

The Monitor Request ID in DECIMAL: 268304394
The (PID.Request#) is 4094.10
Operation is RM_NOT_EQUAL_TO 0 with RM_IMMEDIATE_NOTIFICATION RM_RETURN_NOTIFICATION
Polling Interval is 300 seconds

[Ron Cornwell]

The disk that I am concerned about watching is c4t0d4, but this is actually a LUN on the VA7400 which does not show up as a disk in EMS.

disk 7 0/4/0/0.1.16.0.0.0.4 sdisk CLAIMED DEVICE HP A6189A
/dev/dsk/c4t0d4 /dev/rdsk/c4t0d4

[Christopher McCray_1]

Hello,

Could you please post the portion of your package.conf file referencing the resource you are trying to monitor?


Did you have the cluster down when trying to implement this (required)?

Hope this helps

Chris

[Vince Inman]

You will need to set EMS up to monitor the silkworm. Prior to which you will have to get the switch on the network with an IP address assigned to it.

Once that connectivity is established you can turn EMS monitoring on for the switches and set the parameters necessary for monitoring.

You will have to enable the monitors for the switch in SAM (that's the easiest way).