Hi,
I have recently acquired responsibility for a Digital Unix 4.0D, AlphaServer
4000 5/466.
It has 1.5GB of memory and running a progress database.
What I have noticed is that is has high Paging In activity. It starts to page
in rapidly and then after about 20 seconds it is not so bad, but then within
the next 20 seconds, it starts to page in rapidly again. This cycle is
endlessly repeated. Here is a sample of vmstat
Virtual Memory Statistics: (pagesize = 8192)
procs memory pages intr cpu
r w u act free wire fault cow zero react pin pout in sy cs us sy id
11449 34 161K 167 30K 5627 1640 1887 0 1655 0 268 12K 1K 71 29 0
7456 32 161K 144 30K 13K 3991 4193 0 4094 0 237 9K 1K 60 40 0
4459 32 161K 157 30K 15K 4480 5168 0 4666 0 276 9K 1K 59 41 0
9453 32 161K 241 29K 8785 2593 2906 0 2786 0 161 12K 1K 64 36 0
12457 34 161K 160 30K 7033 1988 1971 0 1972 0 253 7K 1K 60 40 0
10454 34 161K 76 30K 2798 744 1039 0 774 0 197 6K 1K 78 22 0
10455 33 161K 109 30K 1346 281 731 0 345 0 190 5K 1K 83 16 1
11453 34 161K 113 30K 1016 183 643 0 270 0 401 6K 2K 82 18 0
10453 35 161K 128 30K 1012 183 641 0 270 0 163 6K 1K 85 15 0
13455 33 161K 119 30K 1931 408 853 0 508 0 216 7K 1K 80 19 1
9458 32 161K 156 30K 11K 3521 3533 0 3429 0 163 10K 1K 58 41 0
11455 32 161K 167 30K 18K 5570 5537 0 5325 0 391 12K 2K 57 43 0
9456 32 161K 147 30K 11K 3579 3731 0 3536 0 300 9K 1K 60 40 0
9456 33 161K 125 30K 5939 1732 1812 0 1790 0 194 8K 1K 60 40 0
9458 34 161K 127 30K 2932 686 1097 0 746 0 221 8K 1K 80 20 0
5464 32 161K 121 30K 1337 281 737 0 345 0 206 5K 1K 83 17 0
11456 33 161K 128 30K 1341 281 739 0 345 0 385 6K 1K 80 20 1
6462 32 161K 152 30K 1335 266 733 0 345 0 213 5K 1K 84 16 0
7462 33 161K 101 30K 1516 329 783 0 393 0 244 7K 1K 78 22 0
7459 34 161K 236 30K 11K 3380 3276 0 3318 0 119 12K 1K 56 44 0
9461 32 161K 149 30K 16K 5019 5186 0 5135 0 133 8K 1K 60 40 0
Can someone shed some light onto this problem? Are there any kernel parameters
that need tuning or does it need a memory boost?
The sys_check utility recommended that the following kernel parameters be
changed
msgbuf_size and message-buffer-size changed from 4096 to 16384
vm-ubcseqstartpercent changed from 50 to 30
vm-page-free-target changed from 128 to 512
name-cache-hash-size changed from 256 to 16384
vm-mapentries changed from 200 to 16384
vm-vpagemax changed from 16384 to 131072
Any help would be greatly appreciated.
Regards,
Zoran.
Received on Sun Jun 25 2000 - 23:47:24 NZST