Mike, > As "meaningless" as iowait is/was, there was significant value once we > got beyond "the cpu is busy in iowait... add more boards!" Getting to > that point paid a couple of semesters of tuition for my salesman's > daughters.
Iowait is an overloaded term in the Solaris kernel, so the confusion is understandable. The functionality that I was supposed to remove was the calculation of the percentage of CPU time spent waiting for I/O. This is somewhat different from the number of threads that are blocked waiting for I/O. So, the mistake I made was that I accidentally removed some of the logic that exported the number of threads waiting for I/O to userland. > Having one value as a quick gage of "is the storage keeping up with > the server" is very valuable in my environment - not everyone that > looks a performance stats is well-equipped to make use of them (see > tuition statement above). However, leaving these same people blind is > not a good approach either. We have long-term history of vmstat data > and would prefer to continue to use this as an indicator. Yes, and this removal of the number of threads waiting for I/O is a functional regression that I'm obligated to fix. I apologize if this mistake has caused you any inconvenience. > It sounds like the fix for this bug would be a kernel patch - is this > correct? I'm not the expert on such topics, but based upon my understanding this would be correct. > Thank you very much for the quick analysis. You're quite welcome. Thanks for finding this problem. -j _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
