How to setup or enable tracing on a Checkpoint

Technote (FAQ)

Question

A checkpoint is taking too long. A checkpoint is stalled. Checkpoint performance is down.
How do you setup or enable checkpoint tracing?

Answer

1. Set the environment variable TRACECKPT to 1.

export TRACECKPT=1; 
setenv TRACECKPT 1;

2. Restart the database server. 

3. Look for checkpoint tracing information in the message log ( online.log ). Here are several examples of what you could see in the message log: 

First example: 

15:58:58  2040 buffers dirty 
15:58:58  oldest lsn loguniq 1, logpos 0xf8 
15:58:58  Reallocated 1500 dskptrs. 
15:58:58  Reallocated 2000 dskptrs. 
15:58:58  Reallocated 2500 dskptrs. 
15:58:58  2040 dirty pages are to be flushed 

15:59:02  dskflush() took 4 seconds 
15:59:02  wait4critex() took 0 seconds 
15:59:02  N = 2500, Avg = 1260, D_avg = 5040, Dirty = 2040 
15:59:02  Reallocated 2000 dskptrs. 
15:59:02  0 buffers dirty 
15:59:02  oldest lsn loguniq 1, logpos 0xf8 
15:59:02  safe_dskflush() took 0 seconds 
15:59:02  Checkpoint Completed:  duration was 4 seconds.

Second example: 

15:59:56  1 buffers dirty 
15:59:56  oldest lsn loguniq 4, logpos 0x187018 
15:59:56  1 dirty pages are to be flushed 

15:59:56  dskflush() took 0 seconds 
15:59:56  wait4critex() took 0 seconds 
15:59:56  N = 2000, Avg = 945, D_avg = 3781, Dirty = 1 
15:59:56  Reallocated 1500 dskptrs. 
15:59:56  0 buffers dirty 
15:59:56  oldest lsn loguniq 4, logpos 0x187018 
15:59:56  safe_dskflush() took 0 seconds 
15:59:56  Checkpoint Completed:  duration was 0 seconds.

http://www-01.ibm.com/support/docview.wss?uid=swg21323236