Hyperfabric problem

[Manuel Sanmartin]

I have a oracle 9iRAC (2 nodes) connected
with two hyperfabric adapters in each node.

I have configured the transparent local failover and oracle is working over HMP.

The version of Hyperfabric is:

HyprFabrc-00 B.11.23.01

The kernel sends a message :

Hyperfabric: send wait timeout large/negative. Original Value 0x1dcbde60. Corrected to 0x2000000. Process context 'oracle'. File: '/ux/kern/kisu/HFLF/src/common/lowfat/sys_send_wait.c' Line: '795'
Timeout called with negative time.
function == 0xE0000000004E0E90, arg == 0xE00000014E665100, ticks == 0xFFFFFC55, flags == 0x2

I don't see any errors in the oracle logs.

I don't know if this message is problem or
a info.

Thanks you.

[Victor Fridyev]

Hi,

Could you please compare time on both nodes. I'm afraid that there is a significant time difference >10 seconds.
If this is true, try to syncronize the computers by NTP.

[Manuel Sanmartin]

There is a diffence of 2-3 seconds

I will put ntp.

Thanks you.

[FLX]

Hi,

A question:

Your 2 hyperfabric links are they configured as point-to-point or is there a switch between?

HP said :
like if cards were configured for HMP local failover, but this is not supported in point-to-point config, you need a switch for that.


I ask this because our 2 hyperfabric between 2 nodes are point to point and we have problems when one node crashed.

[Manuel Sanmartin]

It's configure point-to-point with
failover without a switch.

It's works for me.

I solve the problem of the message of
the kernel, installing xntp and
sync with a externel ntp server.
Sync the two nodes with xntp doesn't work for me.

[Prashant Zanwar_4]

It can be problem related to some patches which need to be there in your system. Search on all latest patches and issues related to hyperfabric. I strongly see prolem shall some patch rersolution on HP side..

Hope it helps
Thanks
Prashant