Extremly HIGH IO load on Parthenon node

[parthenon]

There is again 100% busy Hard disc system on Parthenon node (this was on that node before July 2006, but was repared after). After such HDD load there are a lot of sequences: all large databases files are corrupted, mail received by sendmail are broken and so on. Support never see any problems, as a rule.
What is it?

[root@vps ~]# date --utc
Sun Oct 8 03:39:04 UTC 2006

[root@vps ~]# vmstat 5 55
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
0 1 32888 89384 463784 4502108 0 0 0 0 0 141 43 6 37 15
0 1 32888 80312 464576 4510676 0 0 0 0 0 1712 0 0 0 100
0 0 32888 79208 465396 4516616 0 0 0 0 0 1614 0 0 0 100
0 0 32888 74664 467344 4521688 0 0 0 0 0 1943 0 0 0 100
0 0 32888 72568 470292 4525240 0 0 0 0 0 2440 0 0 0 100
0 0 32888 63656 472248 4530304 0 0 0 0 0 1824 0 0 0 100
0 0 32888 56864 473332 4537800 0 0 0 0 0 2388 0 0 0 100
0 1 32888 40336 473676 4548376 0 0 0 0 0 1812 0 0 0 100
0 0 32888 28568 474244 4556908 0 0 0 0 0 2394 0 0 0 100
0 1 32888 22584 474816 4565696 0 0 0 0 0 2402 0 0 0 99
0 0 32888 10832 474948 4566604 0 0 0 0 0 1420 0 0 0 100
0 0 32888 14800 474872 4566940 0 0 0 0 0 1640 0 0 0 100
0 1 32888 14768 475224 4572764 0 0 0 0 0 1622 0 1 0 99
0 2 32888 12112 475024 4563864 0 0 0 0 0 1947 0 0 0 99
0 0 32888 19816 475212 4556916 0 0 0 0 0 4225 0 0 0 100
0 1 32888 11720 475164 4551472 0 0 0 0 0 2484 0 0 0 100

[anigmad]

i've had the same problems with high i/o load...even though the vps shows low utilization...support is never helpful in this case. they just say that everything is great or that they've found an offender and fixed the problem..it's like a canned response.