When running jobs where some operations are taking very long, the calculation of the latency percentiles seems to get capped at 16.7s or rather exactly at 2^24-2^16 microseconds. Looking at the code, I'm not sure whether this is intentional or not, the long comment at the beginning of stat.h seems to indicate that the highest bucket should be uncapped, but maybe I am reading this wrong. If it is indeed intentional, then there are two additional questions: 1. Can the code generate a warning when it notices that it reaches its limits? 2. Can I increase the limit by just setting FIO_IO_U_PLAT_BITS to some higher value like 7 or 8? Or do some of the other constants need to be changed accordingly? Yours, Jens -- To unsubscribe from this list: send the line "unsubscribe fio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
