From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; Linux i686; U) Opera 7.54 [en] Description of problem: Notice how the kB_read column never changes? Is this some sort of limit being reached? Somehow I doubt that reading has stopped on that partition. iostat -dk Device: tps kB_read/s kB_wrtn/s kB_read kB_wrtn ------------ /dev/sdc 221.87 1763.42 726.04 1073741823 442086076 /dev/sdc1 221.87 1763.42 726.04 1073741823 442086076 ------------ /dev/sdc 221.87 1763.38 726.03 1073741823 442086664 /dev/sdc1 221.87 1763.38 726.03 1073741823 442086664 ------------ /dev/sdc 221.87 1763.33 726.01 1073741823 442087740 /dev/sdc1 221.87 1763.33 726.01 1073741823 442087740 ------------ kernel : 2.4.21-20.ELsmp iostat -v produces : sysstat version 4.0.7 Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. do a lot of disk transfers 2. run iostat -dk Actual Results: the kB_read field never changes once the threshold is reached. I bet that the other fields have the same limit. Expected Results: the real kb transfer number should be shown (or maybe the counter should have reset?) Additional info:
compiled sysstat-5.0.6 on this machine and it seems to be fixed for that version. Not sure what version it was fixed in, but it seems that making an upgrade of systat available would make this problem go away. sysstat version 5.0.6 Device: tps kB_read/s kB_wrtn/s kB_read kB_wrtn -------------------- sdc 213.18 2092.94 679.92 1728492409 561521144 sdc1 213.18 2092.94 679.92 1728491909 561521144 -------------------- sdc 213.18 2092.93 679.91 1728493077 561521192 sdc1 213.18 2092.93 679.91 1728492577 561521192 -------------------- sdc 213.18 2092.86 679.89 1728493557 561522004 sdc1 213.18 2092.86 679.89 1728493057 561522004
we've put 5.0.5 in the update due shortly... Preview my dev version at: http://people.redhat.com/ccb/sysstat/RHEL3/5.0.5-7.rhel3/
Thanks for the update. I will check the newly updated sysstat on that machine when it becomes available in the regular update channels. As long as 5.0.5 already has this fix (as in 5.0.6), it should be fine.