Description of problem:

A system running 5.4 on IBM HS21 blade have shown very high load average when running the uptime command

# uptime
 17:14:43 up 160 days, 23:59,  1 user,  load average: 323.36, 377.13, 380.51
 
# cat /sys/devices/system/clocksource/clocksource0/current_clocksource
hpet

# cat /proc/stat
cpu  16229967 456 61793497 11035602195 11922623 18219 589415 0
cpu0 2698243 5 4340664 1383290728 332387 0 107718 0
cpu1 1883832 92 39719424 1348931718 171137 0 63264 0
cpu2 2028067 18 2308587 1377117093 9211014 12374 92119 0
cpu3 2262698 117 4603141 1382465550 1362039 13 75970 0
cpu4 1325526 10 3247393 1385890325 265532 9 40899 0
cpu5 2052778 97 2712480 1385732214 213902 2396 55598 0
cpu6 2449563 8 3077661 1385030621 139522 0 72283 0
cpu7 1529255 104 1784143 1387143941 227087 3424 81560 0
intr 1252246936 1025705347 0 0 24 3 0 0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 43 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 115972 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 87329789 0 0 0 0 0 0 35626304 69728673 0 0 0 0 0 0 33740757 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 19 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
ctxt 84918961251
btime 1332252904
processes 5162019
procs_running 1
procs_blocked 0

# cat /proc/loadavg
788.69 487.48 417.30 1/1844 30228

# cat /proc/uptime
13910425.33 953319.39

###  The NTP is running normal =and no synchronization issue

# top -b -n 1 | awk '{if (NR <=7) print; else if ($8 =="D") {print;  count++}} END {print "Total status D: "count}'
top - 17:25:10 up 161 days, 10 min,  1 user,  load average: 788.69, 897.27, 870.44
Tasks: 179 total,   1 running, 178 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.1%us,  0.6%sy,  0.0%ni, 99.2%id,  0.1%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   3997492k total,  1272408k used,  2725084k free,   319660k buffers
Swap:  4209020k total,        0k used,  4209020k free,   470400k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
Total status D:
# df -k
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/sda6             10863172   1219800   9091540  12% /
/dev/sda5             58831072    267576  55526812   1% /data
/dev/sda7             27173752   3173076  22620292  13% /home
tmpfs                  1998744         0   1998744   0% /dev/shm


Version-Release number of selected component (if applicable):


How reproducible:

Not exactly sure how to reproduc the problem. 

Steps to Reproduce:
1.
2.
3.
  
Actual results:


Expected results:


Additional info: