Re: [Freeipmi-users] "Connection timed out" error

[Al Chu]

Hey Dave,

Sorry that FreeIPMI couldn't be of use for you on the HP's.  Both
Ipmitool and FreeIPMI have had to struggle with these random non-
compliance issues from vendors.

I don't think we can add 2.0 session support in a few weeks.  It's
probably going to have to be for our 0.5.0 release when we re-architect
libfreeipmi.  Sorry :-(

If you have a support contract with HP, perhaps you can submit a ticket
with them on this problem.  In order for Ipmi to be better across
vendors, I think its the users that have to complain and let them know
this is not acceptable.  Not much can really be done when vendors only
want to support their vendor provided software against what should be a
industry standard.

Thanks,
Al

On Thu, 2007-05-24 at 10:40 -0400, Davidson Mike wrote:
> Al,
> 
> Ok, thanks for the response.  Would probably be good to support  
> either v1.5 or v2.0
> session establishment depending upon what interface option is  
> specified on the
> command line (similar to IPMItools "lan" and "lanplus" interface  
> options).
> 
> I'm probably going to have to choose a tool fairly soon (within next  
> couple of weeks)
> so most likely I'll either have to extend the ipmi-sensor and ipmi- 
> sel commands to do
> v2.0 session establishment or I'll have to go with ipmitool.
> 
> Thanks for your help.
> 
> Mike
> 
> On May 23, 2007, at 5:40 PM, Al Chu wrote:
> 
> > Hey Dave,
> >
> >> So my question is, does freeIPMI support IPMIv2 session
> >> establishment?  If not, is there
> >> any plan to support it in the future?
> >
> > Ipmipower and ipmiconsole currently support the v2 session  
> > establishment
> > (ipmiconsole b/c of Serial-over-LAN and ipmipower b/c that was a  
> > "test"
> > ground for IPMI 2.0 before ipmiconsole).  We had not planned to  
> > support
> > the IPMI 2.0 session establishment in most of the tools b/c it  
> > shouldn't
> > have been necessary.  According to 13.4 of the specification:
> >
> > "Any IPMI v2.0 conformant BMC ... that supports RMCP+ [i.e. ipmi 2.0]
> > for IPMI messaging and standard payloads is required to simultaneously
> > support IPMI v1.5 packet formats."
> >
> > It's certainly disappointing that HP ignores the above.
> >
> > I have put down that IPMI 2.0 session establishment will be needed in
> > all the tools for the future.  I'm not quite sure when we would be  
> > able
> > to get to it though.  Do you have a deadline for a need?
> >
> > Al
> >
> > On Wed, 2007-05-23 at 15:39 -0400, Davidson Mike wrote:
> >> Al,
> >>
> >> Unfortunately, setting the privilege level did not help.  I used the
> >> tcpdump utility along
> >> with wireshark (formerly ethereal) to analyze the packet exchange
> >> between the linux
> >> box running freeipmi and the target IPMI v2 compliant system.  It
> >> appears that the
> >> 'ipmi-sensors' command is attempting IPMI v1.5 session establishment
> >> which the HP
> >> Proliant server (apparently) does not support.  So the packet
> >> exchange looks something like:
> >>
> >> source                     protocol                packet description      
> >>                                 destination
> >> --------------------------------------------------------------------- 
> >> ---
> >> ----------------------------------------------
> >> freeipmi   box             ASF             Presence Ping                   
> >>                         HP Server
> >> HP Server          ASF             Presence Pong                           
> >>                 freeipmi box
> >> freeipmi box               IPMI            Get Channel Auth Capab. Request 
> >>         HP Server
> >> HP Server          IPMI            Get Channel Auth Capab. Response        
> >> freeipmi box
> >> freeipmi box               IPMI            Get Session Challenge Request   
> >>         HP Server
> >> <no response from HP Server>
> >> freeipmi box               IPMI            Get Session Challenge Request   
> >>         HP Server
> >> <no response from HP Server>
> >> freeipmi box               IPMI            Get Session Challenge Request   
> >>         HP Server
> >> <no response from HP Server>
> >>
> >> impi-sensor command finally gives up and errors out with "connection
> >> timed out" error
> >> I describe in my first email.
> >>
> >> I can reproduce this exact behavior with ipmitool if I issue the
> >> sensor command using the
> >> "lan" interface instead of the new "lanplus" interface option which
> >> supports IPMIv2/RMCP+
> >> session establishment.
> >>
> >> So my question is, does freeIPMI support IPMIv2 session
> >> establishment?  If not, is there
> >> any plan to support it in the future?
> >>
> >> Thanks,
> >> Mike
> >>
> >>
> >>
> >>
> >>
> >> On May 22, 2007, at 9:50 PM, Albert Chu wrote:
> >>
> >>> Hi Dave,
> >>>
> >>> Looking at the ipmitool manpage really quickly, they default to
> >>> using the
> >>> 'admin' privilege for most commands while many of the freeipmi tools
> >>> default to the 'user' privilege (since some tools like ipmi-sensors,
> >>> ipmi-sel do not require a higher privilege).
> >>>
> >>> I'm guessing that's the problem.  Could you perhaps give that a  
> >>> shot?
> >>>
> >>> Thanks,
> >>> Al
> >>>
> >>>> I'm a newbie to the world of IPMI and remote management and am  
> >>>> trying
> >>>> to evaluate FreeIPMI vs ipmitool.  I've downloaded, built and
> >>>> installed FreeIPMI v0.3.3 and ipmitool v1.8.9 on a Fedora Core 5
> >>>> linux box.  I've been able to successfully query an HP Proliant DL
> >>>> 145 server (which supports IPMI 2.0) for a variety of information
> >>>> over the server's BMC LAN interface using ipmitool.  However, when
> >>>> using the FreeIPMI tools (such as ipmi-sensors) I keep getting the
> >>>> following error:
> >>>>
> >>>> $ sudo ipmi-sensors -D lan -h 192.168.43.133 -u admin -p admin
> >>>> ipmi_open_outofband(): Connection timed out
> >>>>
> >>>> I've tried using the '-vv' verbose option but I'm not seeing  
> >>>> anything
> >>>> useful.
> >>>>
> >>>> Just for comparison, here's the result when I use ipmitool:
> >>>>
> >>>> $ ipmitool -I lanplus -H 192.168.43.133 -U admin -P admin sensor
> >>>> ACPI STATE       | 0x0        | discrete   | 0x0100| na        |
> >>>> na        | na        | na        | na        | na
> >>>> CPU0 Pres        | 0x1        | discrete   | 0x0200| na        |
> >>>> na        | na        | na        | na        | na
> >>>> CPU1 Pres        | 0x0        | discrete   | 0x0100| na        |
> >>>> na        | na        | na        | na        | na
> >>>> CPU0 TEMP        | 35.000     | degrees C  | ok    | na        |
> >>>> 10.000    | na        | 66.000    | 68.000    | na
> >>>> CPU1 TEMP        | na         | degrees C  | na    | na        |
> >>>> 10.000    | na        | 66.000    | 68.000    | na
> >>>> SYS TEMP         | 23.000     | degrees C  | ok    | na        |
> >>>> 10.000    | na        | 43.000    | 45.000    | na
> >>>> CPU FAN1         | 6000.000   | RPM        | ok    | na        |
> >>>> 3500.000  | 4000.000  | na        | na        | na
> >>>> CPU FAN2         | 6000.000   | RPM        | ok    | na        |
> >>>> 3500.000  | 4000.000  | na        | na        | na
> >>>> CPU FAN3         | 6000.000   | RPM        | ok    | na        |
> >>>> 3500.000  | 4000.000  | na        | na        | na
> >>>> ....
> >>>>
> >>>>
> >>>> Not sure what I'm doing wrong...I've been through the man pages for
> >>>> the various FreeIPMI utilities and nothing is jumping out at  
> >>>> me.  If
> >>>> anyone can shed any light on what to try next I'd very much
> >>>> appreciate it.
> >>>>
> >>>> Btw, the 'ipmipower', 'ipmiping' and 'rmcpping' utilities all
> >>>> work...I'm able to remotely power the machine on and off.   So I'm
> >>>> guessing this might be some kind of configuration/authentication
> >>>> issue since only the commands that don't require credentials appear
> >>>> to work.  May just be a conincidence.
> >>>>
> >>>> Thanks in advance.
> >>>>
> >>>> Mike
> >>>>
> >>>>
> >>>> _______________________________________________
> >>>> Freeipmi-users mailing list
> >>>> address@hidden
> >>>> http://lists.gnu.org/mailman/listinfo/freeipmi-users
> >>>>
> >>>
> >>>
> >>> -- 
> >>> Albert Chu
> >>> address@hidden
> >>> 925-422-5311
> >>> Computer Scientist
> >>> High Performance Systems Division
> >>> Lawrence Livermore National Laboratory
> >>
> >>
> >>
> >> _______________________________________________
> >> Freeipmi-users mailing list
> >> address@hidden
> >> http://lists.gnu.org/mailman/listinfo/freeipmi-users
> > -- 
> > Albert Chu
> > address@hidden
> > 925-422-5311
> > Computer Scientist
> > High Performance Systems Division
> > Lawrence Livermore National Laboratory
-- 
Albert Chu
address@hidden
925-422-5311
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory