Our new F740 has a pretty constant load (`sysstat 10`) thusly:
CPU NFS CIFS HTTP Net kB/s Disk kB/s Tape kB/s Cache in out read write read write age 46% 1114 1131 0 512 1331 701 371 0 0 5 53% 1559 1316 0 595 1615 446 387 0 0 6 50% 1343 1275 0 596 1211 489 464 0 0 5 48% 1454 1283 0 544 1272 276 395 0 0 6 52% 1844 1275 0 651 1837 280 320 0 0 6 58% 1399 1551 0 1027 2787 1037 450 0 0 6 56% 1796 1351 0 763 1541 352 378 0 0 6 45% 1024 1349 0 620 1422 330 542 0 0 6 58% 1657 1353 0 1193 2991 620 754 0 0 6 51% 1171 1467 0 598 1424 272 453 0 0 6
with occasional (every 5-10 minutes) sustained peaks of:
77% 1242 1201 0 489 1731 5285 5049 0 0 6 81% 1497 1279 0 600 2048 4327 5026 0 0 5 78% 1495 1221 0 563 1589 4287 5488 0 0 4 76% 1229 1213 0 472 1795 4546 5262 0 0 3 81% 1467 1150 0 1345 2634 3683 5746 0 0 2 80% 969 1068 0 397 2051 3876 6298 0 0 2 91% 1049 905 0 384 1102 1230 10622 0 0 2 92% 1328 942 0 448 1801 1536 10336 0 0 2
Are there any rules of thumb about when I should start worrying? I've got only about 2/3 of the data on it and the same fraction of clients connected to it that I need. The low cache numbers already worry me, but it's got the stated max of 1/2 GB.
Max CPU and nfs-ops/sec I've seen is:
86% 3148 563 0 4228 5598 4131 3845 0 0 1 93% 4425 452 0 3573 4503 3283 3977 0 0 1 92% 3320 437 0 6299 4711 3778 5573 0 0 1 95% 3827 552 0 5282 5195 4094 5663 0 0 1
at which point I imagine the robot from "Lost In Space" coming out flailing its arms, if there existed the option "danger.will_robinson on".
(Basic config is two 7 drive shelves of 18 GB FC-AL disks in a single RAID group with one spare and one volume and a quad 100 Mb/s board trunked into a 400 Mb/s fd pipe to the switch. I think I overspec'ed the network connection, though I have seen Net kB in and out burst and exceed 10,000 each, almost the limit of a single 100 Mb/s pipe).
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 ---