From: <rw...@ea...> - 2001-11-02 12:29:13
|
The mmstress test looks like a great addition to LTP. This is the first LTP test I've seen take the load average over 32. I like that. BTW, interactive response is still decent, considering the load. mmstress is only in cvs, so I understand it's not for general use yet, but I have a question. I ran it for a while last night. I'm not sure ran for more than an hour or not. 2.4.14-pre7 came out so I stopped mmstress and booted the new kernel. I started runalltests.sh again and went to bed. ioctl02 hung before runalltests.sh got to mmstress. (ioctl02 hanging has been a very intermittant problem). I killed the 2 ioctl02 processes, so runalltests.sh has moved into mmstress. I'll know tonight when I check the test if mmstress has moved passed test1. I'm hoping in the meantime, someone more knowledgeable will provide some insight. My question is, does mmstress actually exit when no args are passed? Looking in mtest05/mmstress.c, I see: int test_time = 0; /* duration for which test is to be run */ I see that test_time can be changed by passing -t {hours}. I'm wondering if -t isn't passed, whether the value of test_time = 0 ends up becoming a "run forever" test. Thanks for all the good work you do! I'm going to try initializing test_time = 1, as well as letting the current mmstress test run. -- Randy Hron |