freeipmi-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Freeipmi-users] bmc-watchdog problem (UNCLASSIFIED)


From: Hank Bruning
Subject: Re: [Freeipmi-users] bmc-watchdog problem (UNCLASSIFIED)
Date: Mon, 25 Nov 2013 18:11:22 -0500

Someone with more background on the implementation of freeipmi may have
some insight but the busy BMC problem is not specific to freeipmi.

All IPMI based system management software can overload the BMC or the IPMB
bus that is handling the traffic to the nodes.
That is why some IPMI software implement priority schemes for IPMI
messaging (See the graph at the bottom of
http://www.jblade.com/JbDoc2/products/hemi/hemiDC/doc/programmers/HemiTransactions.html)
with a feed back loop from the BMC to sense when it is getting busy.

Software updates that increase the IPMI messaging rate can overload the BMC
and/or the IPMB bus or buses as in ATCA hardware.

You get what you pay for here. The cheaper the BMC is, the more likely it
will be be busy. The expensive BMCs, like Intel's CMM, with parallel IPMB
buses  are more more difficult to overload.

The opposite approach, which is used by large data centers, is to make a
very  cheap BMC without any SDRs, FRU Info records or SEL and move that
intelligence into the remote System Manager. The BMC becomes little more
than an ethernet to IPMB bus translator.

Hank Bruning
JBlade


On Mon, Nov 25, 2013 at 5:32 PM, Hartlieb, George <
address@hidden> wrote:

> Classification: UNCLASSIFIED
> Caveats: NONE
>
> Hi,
>
> I an getting the following error for bmc-watchdog version
> freeipmi-bmc-watchdog-1.3.3-1.el6.x86_64,
> which causes a reboot.
>
> System:
>
> Supermicro IPMI
> CentOS 6.4
> Cluster with 16 nodes
>
> The error happens on all nodes over time, but seems random.
>
> I was running bmc-watchdog 1.0.9 on the same hardware under Centos 5.4
> and never had a problem.
>
>
> /var/log/messages
> /usr/sbin/bmc-watchdog[2620]: Get Cmd: BMC Timeout: driver busy
>
> Nov 25 12:51:08 compute-0-10 /usr/sbin/bmc-watchdog[2620]: fiid_obj_get:
> 'present_countdown_value': data not available
>
>
> Thanks,
>   George
>
>
>
> Classification: UNCLASSIFIED
> Caveats: NONE
>
>
>
> _______________________________________________
> Freeipmi-users mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/freeipmi-users
>


reply via email to

[Prev in Thread] Current Thread [Next in Thread]