Re: IOMETER for Linux network measurement ... "Grunt thread exiting with 1 still active" errors

Was any progress made on #1244848?   Cheung Ming said he would follow up.  I think we are seeing the same problem.

Thank you,

Howard David

Intel Corp.

[ 1244848 ] *** Grunt thread exiting with 1 still active

Date: 2005-08-16 02:19

Sender: cheungming

Logged In: YES 

user_id=896657

this is caused by the bug fix on #1088486. the reason is

that a network io is differ from disk io. for diskio, the

worker stops and then wait current io ends and quits, which

is pretty reasonable. but for network worker, it is possible

that there is always 1 io unfinished for certain unknown

reason, then it dead loop there.

Date: 2005-07-27 13:36

Sender: cheungming

Logged In: YES 

user_id=896657

Thanks for reporting. I will check this once get time.

We have been experimenting with IOMETER for Linux, for evaluating the interaction of power management with network traffic performance, and it works until we get into an endless loop of "*** Grunt thread exiting with 1 still active".  It's not a big deal if we get one of those messages when the IOMETER control function stops the test, but if we get hundreds of messages ... the resulting statistics are no good.

Is anyone using IOMETER with Linux for network performance testing?  Do any of you know who did this work most recently, at Intel?  There was an Intel contractor henryx.w.tieman@... back in 2004, but he's not here any more.

I have not found another micro-benchmark that lets you shape the traffic, and observe the BW / average response / maximum response.  If IOMETER has been replaced by something more recent, we would like to know that.

Thank you,

Howard David and Paul Drews

Intel Corp. / Corporate Technology Group / Systems Technology Lab