memory_write time is extremely long sometimes

[yaozhixiong@foxmail_com]

I am running cesm1.2.0. Memory_write time could be seen in file cpl.*** in directory "run". Generally, the interval time between 2 days is short, that can be seen as followed.

(1)
tStamp_write: model date = 10602 0 wall clock = 2014-07-23 09:20:24 avg dt = 30.99 dt = 29.01
memory_write: model date = 10602 0 memory = 1402.02 MB (highwater) 0.00 MB (usage) (pe= 0 comps= cpl ATM LND OCN ICE GLC ROF WAV)
tStamp_write: model date = 10603 0 wall clock = 2014-07-23 09:20:53 avg dt = 30.98 dt = 29.25
memory_write: model date = 10603 0 memory = 1402.02 MB (highwater) 0.00 MB (usage) (pe= 0 comps= cpl ATM LND OCN ICE GLC ROF WAV)
tStamp_write: model date = 10604 0 wall clock = 2014-07-23 09:21:22 avg dt = 30.97 dt = 29.24
memory_write: model date = 10604 0 memory = 1402.02 MB (highwater) 0.00 MB (usage) (pe= 0 comps= cpl ATM LND OCN ICE GLC ROF WAV)

The interval time is extremely long sometimes, although the model is still running. I do not know why does it happen. I always kill the job and resubmit it. Can anyone help me? Thank you.

(2)
tStamp_write: model date = 10122 0 wall clock = 2014-07-23 07:17:24 avg dt = 1273.82 dt = 1350.66
memory_write: model date = 10122 0 memory = 316.09 MB (highwater) 0.00 MB (usage) (pe= 0 comps= cpl ATM LND OCN ICE GLC ROF WAV)
tStamp_write: model date = 10123 0 wall clock = 2014-07-23 07:40:12 avg dt = 1278.11 dt = 1368.05
memory_write: model date = 10123 0 memory = 316.11 MB (highwater) 0.00 MB (usage) (pe= 0 comps= cpl ATM LND OCN ICE GLC ROF WAV)
tStamp_write: model date = 10124 0 wall clock = 2014-07-23 08:02:08 avg dt = 1279.75 dt = 1315.99
memory_write: model date = 10124 0 memory = 316.12 MB (highwater) 0.00 MB (usage) (pe= 0 comps= cpl ATM LND OCN ICE GLC ROF WAV)