Re: [Freeipmi-users] Re: FreeIPMI SDR caching problem.

[Al Chu]

Hey Peter,

I think I figured it out, I introduced a sequence number wrap around bug
in 0.8.1 when I was "cleaning up" some code.  Most motherboards don't
have anywhere near the number of sensors that the Dell ones do, so they
never reach that point.  I'll put up a beta tar.gz tomorrow.

Al

On Thu, 2010-02-11 at 16:44 -0800, Al Chu wrote:
> Hi Peter,
> 
> > First of all let me thank you for the great work that you and other
> > developers are doing, this is truly a very helpful tool.
> 
> You're definitely welcome.
> 
> > I was thinking of submitting this problem through the GNU FreeIPMI 
> > project bugs page but did not see any activity there so decided to 
> > mail you directly. I hope that that is OK.
> 
> No problem, but in the future it's best to e-mail the
> address@hidden list (I'm CCing it now).
> 
> > Now to the problem. When running ipmi-* tools that can generate SDR cache
> > such as ipmimonitoring and ipmi-sensors with driver type (-D LAN_2_0), the
> > SDR cache generation freezes after some record and then gives the following
> > error, 'ipmi_sdr_cache_create: internal IPMI error'.
> 
> That's definitely not good.
> 
> > After the error I can rerun the command with driver type LAN, and
> > everything succeeds. From that point on I can use driver type LAN_2_0 again
> > and everything will work fine. So it looks like that only SDR cache
> > generation does not seem to work with LAN_2_0 driver. Just for testing I
> > ran a similar command to read the SDR data using ipmitool and using
> > ipmitool's equivalent of LAN_2_0 and everything seemed to work fine.
> > 
> > I have observed this behavior on Dell PowerEdge R300, R610 and R710. And at
> > every run the error will be generated after the same SDR record for each
> > platform. Below are the errors for each one of those server types:
> 
> Luckily, I have a PowerEdge R610 and I've reproduced the bug.  I need to
> figure this out, it's definitely odd.  It does appear to be specific to
> the Dell motherboards.  I'm not sure what the issue is.
> 
> Thanks for bringing it to my attention.  I'll look into it and get back
> to you.
> 
> Al
> 
> > 
> > 
> > R300 (DRAC5):
> > 
> > --------------------------------
> > 
> > Caching SDR record 61 of 80 (current record ID 61)
> > 
> > ipmi_sdr_cache_create: internal IPMI error
> > 
> > --------------------------------
> > 
> > 
> > 
> > R610 (iDRAC):
> > 
> > --------------------------------
> > 
> > Caching SDR record 58 of 124 (current record ID 58)
> > 
> > ipmi_sdr_cache_create: internal IPMI error
> > 
> > --------------------------------
> > 
> > 
> > 
> > R710 (iDRAC):
> > 
> > --------------------------------
> > 
> > Caching SDR record 59 of 115 (current record ID 59)
> > 
> > ipmi_sdr_cache_create: internal IPMI error
> > 
> > --------------------------------
> > 
> > 
> > 
> > If you have a time and opportunity to look at this problem is there
> > 
> > anything I can provide you with that will make this task easier?
> 
> On Thu, 2010-02-11 at 15:47 -0800, Peter Bisroev wrote:
> > Hello Albert,
> > 
> > 
> > 
> > My name is Peter Bisroev and I am using the FreeIPMI package to monitor a
> > 
> > number of DELL servers from a scripted environment.
> > 
> > 
> > 
> > First of all let me thank you for the great work that you and other
> > 
> > developers are doing, this is truly a very helpful tool. I was thinking of
> > 
> > submitting this problem through the GNU FreeIPMI project bugs page but did
> > 
> > not see any activity there so decided to mail you directly. I hope that
> > 
> > that is OK.
> > 
> > 
> > 
> > Now to the problem. When running ipmi-* tools that can generate SDR cache
> > 
> > such as ipmimonitoring and ipmi-sensors with driver type (-D LAN_2_0), the
> > 
> > SDR cache generation freezes after some record and then gives the following
> > 
> > error, 'ipmi_sdr_cache_create: internal IPMI error'.
> > 
> > 
> > 
> > After the error I can rerun the command with driver type LAN, and
> > 
> > everything succeeds. From that point on I can use driver type LAN_2_0 again
> > 
> > and everything will work fine. So it looks like that only SDR cache
> > 
> > generation does not seem to work with LAN_2_0 driver. Just for testing I
> > 
> > ran a similar command to read the SDR data using ipmitool and using
> > 
> > ipmitool's equivalent of LAN_2_0 and everything seemed to work fine.
> > 
> > 
> > 
> > I have observed this behavior on Dell PowerEdge R300, R610 and R710. And at
> > 
> > every run the error will be generated after the same SDR record for each
> > 
> > platform. Below are the errors for each one of those server types:
> > 
> > 
> > 
> > R300 (DRAC5):
> > 
> > --------------------------------
> > 
> > Caching SDR record 61 of 80 (current record ID 61)
> > 
> > ipmi_sdr_cache_create: internal IPMI error
> > 
> > --------------------------------
> > 
> > 
> > 
> > R610 (iDRAC):
> > 
> > --------------------------------
> > 
> > Caching SDR record 58 of 124 (current record ID 58)
> > 
> > ipmi_sdr_cache_create: internal IPMI error
> > 
> > --------------------------------
> > 
> > 
> > 
> > R710 (iDRAC):
> > 
> > --------------------------------
> > 
> > Caching SDR record 59 of 115 (current record ID 59)
> > 
> > ipmi_sdr_cache_create: internal IPMI error
> > 
> > --------------------------------
> > 
> > 
> > 
> > If you have a time and opportunity to look at this problem is there
> > 
> > anything I can provide you with that will make this task easier?
> > 
> > 
> > 
> > PS: I would be more than happy to submit a patch for the problem but my
> > 
> > knowledge of IPMI protocol is almost non existent so that is not an option
> > 
> > for the meantime.
> > 
> > 
> > 
> > Thanks you.
> > 
> > 
> > 
> > Best Regards,
> > 
> > Peter
> > 
> > 
-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory