This Blog demonstrates one step-by-step test case.
Note: Tested in Oracle 19c with:
1. Test Steps
Open one UNIX window, and two Sqlplus sessions (SID-1 and SID-2).
1.1 UNIX Window
On UNIX Window, find ckpt process and its associated control files,
set GDB breakpoint when pread64 on them.
$ > ps -efl |grep ckpt
0 S oracle 81749 1 0 80 0 - 1246636 - 11:34 ? 00:00:00 ora_ckpt_testdb01
$ > lsof -p 81749 | grep control
ora_ckpt_ 81749 oracle 256uW REG 8,1 32620544 671095912 /oratestdb01/oradata/testdb01/control01.ctl
ora_ckpt_ 81749 oracle 257uW REG 8,1 32620544 671095913 /oratestdb01/oradata/testdb01/control02.ctl
ora_ckpt_ 81749 oracle 258uW REG 8,1 32620544 671095914 /oratestdb01/oradata/testdb01/control03.ctl
$ > gdb -p 81749
(gdb) display $rdi
1: $rdi = 31
(gdb) break pread64 if $rdi == 256 || $rdi == 257 || $rdi == 258
Breakpoint 1 at 0x7fc984de9870 (2 locations)
(gdb) info b
Num Type Disp Enb Address What
1 breakpoint keep y
stop only if $rdi == 256 || $rdi == 257 || $rdi == 258
1.1 y 0x00007fc984de9870
1.2 y 0x00007fc9855052b0
(gdb) c
Continuing.
Breakpoint 1, 0x00007fc9855052b0 in pread64 () from /lib64/libpthread.so.0
1: $rdi = 256
(gdb) info r
rax 0x1 1
rbx 0x0 0
rcx 0x4000 16384
rdx 0x4000 16384
rsi 0x7fc9827a1000 140503454191616
rdi 0x100 256
1. 2. Sqlplus SID-1
On the first Sqlplus session, run some DML and then force Oracle Database to perform a checkpoint.
CKPT starts to read control files, and stops on pread64 breakpoint.
Oracle v$session shows CKPT event "control file sequential read".
create table ksun_tab as
select level x, rpad('ABC', 500, 'X') y from dual connect by level <= 1e4;
Sqlplus > update ksun_tab set y = rpad('ABC', 500, 'x');
10000 rows updated.
Sqlplus > commit;
Commit complete.
Sqlplus > alter system checkpoint;
ERROR:
ORA-03114: not connected to ORACLE
alter system checkpoint
*
ERROR at line 1:
ORA-03113: end-of-file on communication channel
Process ID: 80361
Session ID: 283 Serial number: 3022
1.3. Sqlplus SID-2
On the second Sqlplus session, force Oracle Database to begin writing to a new redo log file group (log switch),
and Oracle Database also begins to perform a checkpoint, hence LGWR is blocked with Event: "enq: CF - contention":
Sqlplus > alter system switch logfile;
ERROR:
ORA-03114: not connected to ORACLE
alter system switch logfile
*
ERROR at line 1:
ORA-03113: end-of-file on communication channel
Process ID: 81230
Session ID: 24 Serial number: 25591
1.4. Instance is being terminated
After 900 seconds, alert.log shows:
Cause - 'Instance is being terminated due to ospid CKPT holding an enqueue CF'
2. DB alert.log
alert.log shows that CKPT is killed by one Background processes (e.g, LGWR, ARCn, TTnn, MMON) due to
ORA-00494: enqueue [CF] held for too long (more than 900 seconds) by CKPT.
Errors in file /orabin/app/oracle/admin/testdb01/diag/rdbms/testdb01/testdb01/trace/testdb01_mmon_80100.trc (incident=33857):
ORA-00494: enqueue [CF] held for too long (more than 900 seconds) by 'inst 1, osid 80070'
Incident details in: /orabin/app/oracle/admin/testdb01/diag/rdbms/testdb01/testdb01/incident/incdir_33857/testdb01_mmon_80100_i33857.trc
2026-08-25T10:37:19.316424+02:00
Killing enqueue blocker (pid=80070) on resource CF-00000000-00000000-00000000-00000000 by (pid=80100)
by killing session 273.24900
KILL SESSION for sid=(273, 24900):
Reason = RAC enqueue blocker
Mode = KILL SOFT -/-/-
Requestor = MMON (orapid = 32, ospid = 80100, inst = 1)
Owner = Process: CKPT (orapid = 21, ospid = 80070)
Result = ORA-29
Killing enqueue blocker (pid=80070) on resource CF-00000000-00000000-00000000-00000000 by (pid=80100)
by terminating the process
MMON (ospid: 80100): terminating the instance due to ORA error 2103
Cause - 'Instance is being terminated due to ospid 80070 holding an enqueue CF-00000000-00000000-00000000-00000000)'
2026-08-25T10:37:19.372998+02:00
System state dump requested by (instance=1, osid=80100 (MMON)), summary=[abnormal instance termination].
3 Background Incident and Trc files
Dump continued from file: /orabin/app/oracle/admin/testdb01/diag/rdbms/testdb01/testdb01/trace/testdb01_mmon_80100.trc
[TOC00001]
ORA-00494: enqueue [CF] held for too long (more than 900 seconds) by 'inst 1, osid 80070'
[TOC00001-END]
[TOC00002]
========= Dump for incident 33857 (ORA 494) ========
[TOC00003]
----- Beginning of Customized Incident Dump(s) -----
-------------------------------------------------------------------------------
ENQUEUE [CF] HELD FOR TOO LONG
enqueue holder: 'inst 1, osid 80070'
Process 'inst 1, osid 80070' is holding an enqueue for maximum allowed time.
The process will be terminated after collecting some diagnostics.
It is possible the holder may release the enqueue before the termination
is attempted. If so, the process will not be terminated and
this incident will only be served as a warning.
Oracle Support Services triaging information: to find the root-cause, look
at the call stack of process 'inst 1, osid 80070' located below. Ask the
developer that owns the first NON-service layer in the stack to investigate.
Common service layers are enqueues (ksq), latches (ksl), library cache
pins and locks (kgl), and row cache locks (kqr).
[880 samples, 10:35:48 - 10:50:48]
waited for 'control file sequential read', seq_num: 16046
p1: 'file#'=0x0
p2: 'block#'=0x1
p3: 'blocks'=0x1
Problem Key: ORA 494
Error: ORA-494 [[CF]] [900] [1] [80070] [] [] [] [] [] [] [] []
[00]: dbgexExplicitEndInc [diag_dde]
[01]: dbgeEndDDEInvocationImpl [diag_dde]
[02]: dbgeEndSpltInvokOnRec [diag_dde]
[03]: dbgePostErrorKGE [diag_dde]
[04]: dbkePostKGE_kgsf [rdbms_dde]
[05]: kgeade []
[06]: kgeselv []
[07]: ksesec3 [KSE]
[08]: ksdx_cmdreq_wait_for_pending [VOS]<-- Signaling
[09]: ksdxdocmdmultex [VOS]
*Note: 10:35:48 - 10:50:48 is 15 minutes (900 seconds)
bpftrace -v -e 'tracepoint:syscalls:sys_enter_io_submit /strncmp("ora_ckpt", comm, 7) == 0/ {
$iocb_sizeof = sizeof(struct iocb);
$reqs = args->nr;
$iocbpp_ptr = (struct iocb **)args->iocbpp;
time("%H:%M:%S --- ");
printf("%s, iocb_sizeof: %d, io_submit I/O request: %d\n", comm, $iocb_sizeof, $reqs);
$i = 1; while ($i <= $reqs) { $iocbpp_ptr_2 = (struct iocb **)($iocbpp_ptr + $iocb_sizeof*($i-1));
$iocbpp_struct = (struct iocb *)(*$iocbpp_ptr_2);
printf("--- File: %d --- ptr_2 %p, aio_fildes: %d, aio_nbytes: %d, aio_offset: %ld, aio_lio_opcode(1=PWRITE): %d, aio_data:0X%X\n",
$i, $iocbpp_ptr_2, $iocbpp_struct->aio_fildes, $iocbpp_struct->aio_nbytes, $iocbpp_struct->aio_offset, $iocbpp_struct->aio_lio_opcode, $iocbpp_struct->aio_data);
@All_CNT = count(); @CNT[$iocbpp_struct->aio_fildes] = count(); @SUM[$iocbpp_struct->aio_fildes] = sum( $iocbpp_struct->aio_nbytes);
if ($i > 50) { break; }
$i++ }
printf("\n");}'
