OES11 SP1 - sfcb failed to load libhms_class messages

  • 7014360
  • 27-Dec-2013
  • 18-Apr-2014


SUSE Linux Enterprise Server 11 Service Pack 2 (SLES 11 SP2)
Novell Open Enterprise Server 11 (OES 11) Linux


During the initial OES11 SP1 installation, it is possible that a large number of messages are observed that seem to indicate problems with the sfcb service.

Fore example in /var/log/messages repetitive series of the following messages can be seen:

sfcb[16933]: *** Failed to load libhms_class_LinuxPhysicalMemory.so for hms_class_LinuxPhysicalMemory
sfcb[16948]: *** Failed to load libhms_class_AggregateProcessors.so for hms_class_AggregateProcessors
sfcb[16952]: *** Failed to load libhms_class_LinuxPhysicalMemory.so for hms_class_LinuxPhysicalMemory
sfcb[16961]: *** Failed to load libhms_class_LinuxVirtualMemory.so for hms_class_LinuxVirtualMemory
sfcb[16965]: *** Failed to load libhms_class_LinuxPhysicalMemory.so for hms_class_LinuxPhysicalMemory
sfcb[16998]: *** Failed to load libhms_class_NetworkDevice.so for hms_class_NetworkDevice

Another sfcb related problem symptom which was observed at times during OES installation/configuration as well, was a segfault in sfcb.

snippet taken from /var/log/messages below :

kernel: [253128.691218] sfcbd[11538] general protection ip:7f11459c92e2 sp:7fff34814d00 error:0 in libsfcBrokerCore.so.0.0.0[7f11459ab000+57000]
kernel: [253128.697598] sfcbd[11543]: segfault at 100000018 ip 00007f1143be43a3 sp 00007fff34813010 error 4 in libc-2.11.1.so[7f1143b6f000+154000]

*** glibc detected *** sfcbd: free(): invalid pointer: 0x00000000007aea70 ***


Configuring the OES services following the OES installation addresses the  Failed to load libhms_class_** messages problem.
After the OES installation and configuration finishes, this will add required path and location details for the HMS providers to the /etc/sfcb/sfcb.cfg file. Once sfcb is restarted hereafter, these problems are confirmed to be resolved.
Example of what should be seen in /etc/sfcb/sfcb.cfg on an OES installation.
providerDirs: /usr/lib64/sfcb /usr/lib64 /usr/lib64/cmpi /opt/novell/lib64/cmpi /opt/novell/lib64/sfcb/cmpi

The problem of the sfcbd crash was occasionally observed when the service was started, and this problem has been resolved with an sfcb update. 
Once updating the server with the latest available patches from the patch catalogs and restarting the same, this problem is also resolved.


There are several OES services consuming SFCB. Whilst the SFCB service up and running and OES services calling into SFCB whilst still being unconfigured causes these messages to be logged.
Novell Remote Manager starts making calls to SFCB before it has modified the /etc/sfcb/sfcb.cfg file to set the correct path and location of the HMS providers, and as a results the providers do not load.

The segfault was caused when trying to free the same memory twice.