Here's our output, for comparison. Seems to be a little more evenly distributed than yours, but e3a takes the lead by far.
bat> netstat -i Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Collis Queue e0* 1500 none none 0 0 0 0 0 0
e3a 1500 144.212 bat 935492178 17 1070359540 268973 0 0 e3b 1500 144.212 bat 712967233 17 800877896 233521 0 0 e3c 1500 144.212 bat 527507321 2 669191765 212388 0 0 e3d 1500 144.212 bat 522712811 89 672437298 201463 0 0 lo 1536 127 localhost 30198 0 30198 0 0 0
---
How much bandwidth are you using there, I am only barely using 110-130Mbits average.
Im thinking it hasnt reached a 'threshold' to fire up more links.
On Wed, 3 Mar 1999, Mohler, Jeff wrote:
How much bandwidth are you using there, I am only barely using 110-130Mbits average.
Im thinking it hasnt reached a 'threshold' to fire up more links.
The filer is quite busy at the moment:
CPU NFS CIFS HTTP Net kB/s Disk kB/s Tape kB/s Cache in out read write read write age 75% 2695 1322 0 1554 2930 769 0 0 0 3 86% 2596 1150 0 2668 6273 4564 1953 0 0 2 94% 2515 1125 0 4332 6292 6295 3998 0 0 2 92% 2447 1197 0 3581 6829 6164 3836 0 0 2 87% 2257 1017 0 3251 6568 7332 3976 0 0 2 78% 2683 1028 0 1482 7961 5595 0 0 0 1 90% 2796 1057 0 743 10365 5145 3399 0 0 1 88% 2525 1262 0 2772 9479 5426 0 0 0 1
Usually, it's about:
39% 1048 519 0 948 1503 962 2215 0 0 5 40% 1121 765 0 761 1424 485 0 0 0 5 45% 1043 1206 0 393 1996 493 0 0 0 5 52% 1882 843 0 509 2701 1305 1336 0 0 5 45% 2188 653 0 570 2120 256 0 0 0 5 35% 1386 463 0 368 1013 388 795 0 0 5 49% 1789 507 0 733 3110 1656 0 0 0 5 43% 1653 626 0 523 2623 288 673 0 0 5 42% 1840 447 0 588 3422 1878 0 0 0 5 59% 1926 685 0 1414 5285 2088 1320 0 0 4 48% 1959 642 0 1273 3353 473 0 0 0 5
(100 Mb/s ~= 12,200 Kb/s, neglecting packet overhead)
So, most of the time we are well under even a single 100 Mb/s pipe, and are getting some data on all the trunks, just imbalanced. I have seen it peak above 12,200 Kb/s in and out, so even though it's overspeced 95% of the time, we do take advantage of the trunking. The filer is connected to a Cisco WS-X5225R card in a Cat5000 on the other end, if that helps.
What I need desperately is more cache. We had this sucka pegged this morning:
CPU NFS CIFS HTTP Net kB/s Disk kB/s Tape kB/s Cache in out read write read write age 99% 3142 973 0 4352 6359 5978 5802 0 0 0 96% 3440 824 0 4197 7344 6358 4292 0 0 0 98% 3051 1072 0 4105 8944 7733 4458 0 0 0 98% 3193 789 0 4844 10042 9329 3799 0 0 0 99% 3380 1068 0 3210 8335 7419 3828 0 0 0 97% 3223 1064 0 3502 8060 7099 3949 0 0 0 99% 3252 1234 0 3421 8465 7224 2535 0 0 0
"Danger, Will Robinson."
[0] `systat 5` throughout.
Until next time...
Todd C. Merrill The Mathworks, Inc. 508-647-7792 24 Prime Park Way, Natick, MA 01760-1500 508-647-7012 FAX tmerrill@mathworks.com http://www.mathworks.com ---
-
Mohler, Jeff -
Todd C. Merrill