2602773 - Enabling I/O fencing on VMWare may crash server or get signal 11 in decode_dev_ids - SAP ASE CE

• During installation or restarting the ASE Cluster Edition the signal 11 on function decode_dev_ids may reported and even at time the server may crash.

Current process (0x0) infected with signal 11 (SIGSEGV)
Address 0x0x000000000132cbd5 (decode_dev_ids+0x44), siginfo (code, address) = (1, 0x0x00007ffdfbb4c028)

• This is noticed only when enabling “enable i/o fencing” flag on and from OS the I/O fencing is enabled.
   
• In the ASE Errorlog we can notice the below errors

kernel  XXXX module: decode_dev_ids: Warning guid is too long. It will be truncated.
kernel  XXXX module: decode_dev_ids: Warning guid is too long. It will be truncated.
kernel  XXXX module: scsi_pgr_reserve: ioctl failed(Success)
kernel  XXXX module: xxxx_reserve_device: Error(-2) while reserving key(4bc96680)
kernel  XXXX module: xxxx_reserve_device: Unregistering host key(4bc96680)
kernel  iof_reserve_device: Error(-9) while reserving fence device /dev/sg4
kernel  udactivate: error -8 while reserving virtual device 2 for I/O fencing
server  Disk initialization for logical device 'temp_log_dev1' (physical path '/dev/raw/raw5') failed due to I/O fencing error and prevented instance 'ase1' (1) from joining the cluster. Refer to the instance error log for additional information.
kernel  ueshutdown: exiting

• When signal 11 occurs, the following calls are reported in the stack trace:

decode_dev_ids
uc_stack:ss_size:ss_flags
nthrd_backtrace

   Note: More messages (not a complete stack trace) are avaialable under Attachments

• SAP Adaptive Server Enterprise Cluster Edition (ASE CE) 15.7
• SAP Adaptive Server Enterprise Cluster Edition (ASE CE) 16.0
• VM-Ware configuration

SEGV, segmentation, storage, access, violation, fault, segfault, kisignal, stacktrace,  stack, registers, CR#813411, 813411, CR813411, VM-Ware, iofencing, fencing , KBA , BC-SYB-ASE-CE , ASE Cluster Edition (Standalone) , Bug Filed