Andy/Al, Thanks for all your help. It ends up I'm back where I started. A little bit more on what I'm doing. I'm testing with 5 iDrac blades in a Dell m1000e chassis. My program basically uses ipmitool to query an IP range that includes all those BMC's. Some of the IP's in the range don't have ipmi servers listening on the other end (the chassis for instance)
ipmitool works fine the first time through after I've power cycled the servers. The next time, I begin to see resource problems with either lan or lanplus mode. My suspicion is I'm either not closing ipmi connections I need to be closing or I'm not respecting some timing. Since I'm in the order of seconds between ipmi commands, I can't imagine it's a timing issue. I've assumed ipmitool takes care of closing connections but maybe I'm assuming too much? Is there an easy way for me to see if the server is ready for another request or a command to close an ipmi session I'm not aware of? Thanks, Dick -----Original Message----- From: Al Chu [mailto:ch...@llnl.gov] Sent: Monday, December 14, 2009 3:54 PM To: Andy Cress Cc: Detweiler, Dick; ipmitool-devel@lists.sourceforge.net Subject: Re: [Ipmitool-devel] RAKP Insufficient Resources response from aniDRACm610 blade... Hi Dick, Andy, > since all IPMI 2.0 implementations are required to support > the 1.5 LAN protocol as well. While required by the IPMI specification, there is atleast 1 major vendor (for some version of a product line) where this was not true. They supported IPMI 2.0 but not IPMI 1.5. So depending on your app you may want to keep that in mind for potential portability issues. However, it's such a small case, it's probably no big deal if you ignored it. Al On Mon, 2009-12-14 at 13:01 -0800, Andy Cress wrote: > Dick, > > It would seem simpler and more robust to just execute your commands with > the IPMI LAN 1.5 protocol (lan) instead of the IPMI LAN 2.0 protocol > (lanplus), since all IPMI 2.0 implementations are required to support > the 1.5 LAN protocol as well. > > The 1.5 LAN session protocol isn't as complex and shouldn't use up as > many buffers in the target BMC. > > Note that each vendor's BMC implementation may have an internal limit on > the number of simultaneous sessions. > > Andy > > -----Original Message----- > From: Detweiler, Dick [mailto:dick.detwei...@avocent.com] > Sent: Monday, December 14, 2009 3:27 PM > To: ipmitool-devel@lists.sourceforge.net > Subject: [SPAM] - [Ipmitool-devel] RAKP Insufficient Resources response > from an iDRACm610 blade... - Email found in subject > > Hi all, > > I am writing an app using ipmitool (among other things) to find out the > type of blades that are on the network. For each IP address in a given > range, a thread is created so the discovery process is being run > concurrently on many targets. There is a mutex around the calls to the > ipmi code so only one thread is using the ipmi driver at a time. > > The application is running on Linux using version 1.8.10 of ipmitool. > When it uses ipmitool, it first send an asf ping to retrieve the IPMI > level supported on the device. If that is successful, I follow up > sending the following command: > > Ipmitool -I lanplus -L ADMINISTRATOR+ -U user -P pass -H > 192.168.what.ever fru print > > Most of the time I get back the information I want but sometime it > generates an error. The BMC will return error code 1:' insufficient > resources for session'. Here is the output from a -vv invocation of > the above command: > > ======================================= > IPMI LAN host 192.168.96.52 port 623 > > >> Sending IPMI command payload > >> netfn : 0x06 > >> command : 0x38 > >> data : 0x8e 0x04 > > >> SENDING AN OPEN SESSION REQUEST > > >> Console generated random number (16 bytes) > 17 22 e1 cf 4a 96 88 c4 b2 7a 2a 46 95 40 bc 54 > >> SENDING A RAKP 1 MESSAGE > > RAKP 2 message indicates an error : insufficient resources for session > ======================================= > > Questions: > > - I assume this is a resource problem on the target BMC and not with the > ipmitool client on my side. Following the ipmitool code it seems like > the only place that error code can be set is from the data returned by > the BMC. Unless I missed something? > > - If this is a resource issue on the BMC, what might the problem be? > Too fast at getting back to the BMC after the asf ping? Too many > connections open on the BMC? > > Any help is appreciated. > > Thanks, > > Dick Detweiler > > > ------------------------------------------------------------------------ > ------ > Return on Information: > Google Enterprise Search pays you back > Get the facts. > http://*p.sf.net/sfu/google-dev2dev > _______________________________________________ > Ipmitool-devel mailing list > Ipmitool-devel@lists.sourceforge.net > https://*lists.sourceforge.net/lists/listinfo/ipmitool-devel > ourceforge.net/lists/listinfo/ipmitool-devel > > > > ------------------------------------------------------------------------ ------ > Return on Information: > Google Enterprise Search pays you back > Get the facts. > http://*p.sf.net/sfu/google-dev2dev > _______________________________________________ > Ipmitool-devel mailing list > Ipmitool-devel@lists.sourceforge.net > https://*lists.sourceforge.net/lists/listinfo/ipmitool-devel > -- Albert Chu ch...@llnl.gov Computer Scientist High Performance Systems Division Lawrence Livermore National Laboratory ------------------------------------------------------------------------------ Return on Information: Google Enterprise Search pays you back Get the facts. http://p.sf.net/sfu/google-dev2dev _______________________________________________ Ipmitool-devel mailing list Ipmitool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ipmitool-devel
