PI83758: CLEARCASE DATABASE PROCESSES SEGFAULT ON SOLARIS IF DB.CONF FILECONTAINS "-SUPPRESS_DSYNC"

Rational ClearCase Fix Pack 5 (9.0.0.5) for 9.0
Rational ClearCase Fix Pack 1 (9.0.1.1) for 9.0.1
Rational ClearCase Fix Pack 6 (9.0.0.6) for 9.0
Rational ClearCase Fix Pack 2 (9.0.1.2) for 9.0.1
Rational ClearCase Fix Pack 3 (9.0.1.3) for 9.0.1
Rational ClearCase Fix Pack 4 (9.0.1.4) for 9.0.1
Rational ClearCase Fix Pack 5 (9.0.1.5) for 9.0.1
Rational ClearCase Fix Pack 6 (9.0.1.6) for 9.0.1
Rational ClearCase Fix Pack 7 (9.0.1.7) for 9.0.1
Rational ClearCase Fix Pack 8 (9.0.1.8) for 9.0.1
Rational ClearCase Fix Pack 9 (9.0.1.9) for 9.0.1
Rational ClearCase Fix Pack 10 (9.0.1.10) for 9.0.1
Rational ClearCase Fix Pack 11 (9.0.1.11) for 9.0.1
Rational ClearCase Fix Pack 12 (9.0.1.12) for 9.0.1
Rational ClearCase Fix Pack 13 (9.0.1.13) for 9.0.1

APAR status

• Closed as program error.

Error description

• After upgrading to ClearCase 9.0.0.3/4 on Solaris 11.3 hosts,
  vob server processes immediately terminate with segmentation
  faults.
  
  Investigation showed that the failures all had the following
  stack trace:
  % /usr/proc/bin/pstack core.db_server_0_9970
  core 'core.db_server_0_9970' of 9970:	db_server
  -S15:21,tcp:22,tcp
   07595959 fgets    (feffb9d0, 400, 7600040) + 21
   07eea470 get_config_options (feffd2b4, feffd2f8, 80e44bc, 0, 0,
  0) + 700
   07eeac92 open_db  (810f6a0, 810f720, 1, 0, 0, 9) + 36
   07eeb5c4 db_open_db (810f6a0, 810f720, 1, 9, 80ef708) + 20
   08074038 db_open_db_V3 (feffd2b4, feffd2f8, 80f40a0) + 50
   08083c5b dbrpc_3  (feffddf0, 80f40a0) + 42db
   08072652 dbrpc_3_wrap (feffddf0, 80f40a0) + 112
   07bfd140 ks_svc_handle_message (80f40a0, feffd460, feffddf0) +
  e8
   07bfcf9d ks_svc_getreqset (80fa350) + 2c9
   08072b7c client_rpc (80fa350, 0) + 20
   07c35b33 rt_wait  (feffdf38) + 90b
   07c364c2 rt_run   (feffe788, feffe680, fe7db968, 7610bc0, 0, 0)
  + 1de
   08073c2b main     (2, feffe6c4, feffe6d4) + 537
   08071572 _start   (3, feffe7fa, feffe804, feffe805, 0,
  feffe818) + 7a
  
  This issue is currently only observed on Solaris 11.3 hosts.
  
  On investigation, it the "-suppress_dsync" option is being
  parsed differently on 8.0.x and 9.0.x.
  

Local fix

• 1) remove the -suppress_dsync option from the db.conf file. This
  may have performance impacts if ZFS is in use.
  2) change "-suppress_dsync" to "-suppress_dsync " (trailing
  space) in db.conf.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * Users of ClearCase VOBs and the db.conf file.                *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * The ClearCase db_server processes can behave erratically if  *
  * single token lines are placed in the db.conf file.           *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  

Problem conclusion

• A fix is available in ClearCase versions 9.0.0.5 and 9.0.1.1
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  CC&CC MSITE WIN

• Fixed component ID

  5724G3300

Applicable component levels

• R900 PSY

     UP

• R901 PSY

     UP