Closed GoogleCodeExporter closed 9 years ago
stressapptest-1.0.1 is a very old version that only supports up to 32 cores.
Later versions will work fine. The current stressapptest should build, so
please file a bug for that.
Original comment by nsanders@google.com
on 10 Apr 2014 at 6:13
Hi, but we can't build the new version: stressapptest-1.0.6 on CentOS5.2, it
will show make error.
How to resolve it? Thanks
Original comment by ipmit...@gmail.com
on 10 Apr 2014 at 11:50
Attachments:
Please try "./configure --disable-default-optimizations" which will disable
-Wno-psabi
Original comment by nsanders@google.com
on 11 Apr 2014 at 12:22
Hi, it's still show error message when make and make install.
please refer to the attachment file. Thank you
Original comment by ipmit...@gmail.com
on 11 Apr 2014 at 2:09
Attachments:
You may still be using an obsolete version.
Please run the following to get the most up to date code. This should work:
svn checkout http://stressapptest.googlecode.com/svn/trunk/
stressapptest-read-only
Original comment by nsanders@google.com
on 11 Apr 2014 at 11:30
I've removed -Wno-psabi so the current checkout should just build with no
special options.
Original comment by nsanders@google.com
on 11 Apr 2014 at 11:33
which one should I download...?
•Android.mk
•COPYING
•MODULE_LICENSE_APACHE2
•Makefile.am
•Makefile.in
•NOTICE
•aclocal.m4
•config.guess
•config.sub
•configure
•configure.ac
•depcomp
•install-sh
•missing
•src/
•stressapptest.1
Original comment by ipmit...@gmail.com
on 13 Apr 2014 at 9:29
Refer to this page:
https://code.google.com/p/stressapptest/source/checkout
and use svn to download all the files.
Then run
"./configure"
"make"
Original comment by pud...@gmail.com
on 13 Apr 2014 at 9:07
Hi, I have download the stressapptest-read-only and tried to make again.
It's still show make error, please refer to the attachment
file.(readonly_makeerror.jpg)
Original comment by ipmit...@gmail.com
on 14 Apr 2014 at 2:32
Attachments:
sched_getcpu is a glibc function, which was introduced some years ago. However
CentOS 5.2 is from 2008 and predates this glibc. So you may consider updating
your OS version.
See: http://man7.org/linux/man-pages/man3/sched_getcpu.3.html
You can try adding your own implementation of this function at the top of
worker.cc maybe, similar to what's suggested in the manpage? Or simply make an
empty function of that name and have it return zero if you don't care what CPU
each thread is running on.
Original comment by pud...@gmail.com
on 14 Apr 2014 at 11:46
So, can it make normally if I update glibc?
Or, could you help to build the stress test tool for CentOS5.2 usage?
Original comment by ipmit...@gmail.com
on 15 Apr 2014 at 5:53
You can make it normally if you upgrade glibc. However this is probably not
possible without upgrading your whole OS.
Try replacing your file "src/worker.cc" with this one I have attached, and it
may work. I have added a hack to allow it to not use sched_getcpu, maybe it can
compile on your machine.
Original comment by nsanders@google.com
on 15 Apr 2014 at 9:22
Attachments:
Hi, I have updated glibc and this issue still happened, please refer to the
attachment file.
After replaced work file, this is still make error, please refer to
ReplaceWORK.jpg.
Thank you
Original comment by ipmit...@gmail.com
on 16 Apr 2014 at 2:04
Attachments:
Maybe your kernel is too old to support cpu detection? I'm not sure.
Try this file. It just reports every CPU as zero.
Original comment by nsanders@google.com
on 16 Apr 2014 at 4:56
Attachments:
Wow, it's pass for make command.
I'm running the stress test via command: stressapptest -s 259200.
I'm looking forward the pass result, thanks for your help.
Original comment by ipmit...@gmail.com
on 16 Apr 2014 at 5:45
what's your mean about "It just reports every CPU as zero" ?
Have any effect in my test?
Original comment by ipmit...@gmail.com
on 16 Apr 2014 at 11:53
After stress 72hr, the screen show fail message:"test discovered HW problems"
below:
What is the Fail message? How to check detail message?
2014/04/19-13:41:45(CST) Log: Thread 1 found 1 hardware incidents
2014/04/19-13:41:45(CST) Log: Thread 2 found 3 hardware incidents
2014/04/19-13:41:45(CST) Log: Thread 8 found 1 hardware incidents
2014/04/19-13:41:45(CST) Log: Thread 21 found 3 hardware incidents
2014/04/19-13:41:45(CST) Log: Thread 37 found 2 hardware incidents
2014/04/19-13:41:45(CST) Log: Thread 40 found 2 hardware incidents
2014/04/19-13:41:45(CST) Stats: Found 12 hardware incidents
2014/04/19-13:41:45(CST) Stats: Completed: 8230056960.00M in 259201.08s
31751.63MB/s, with 12 hardware incidents, 0 errors
2014/04/19-13:41:45(CST) Stats: Memory Copy: 8230056960.00M at 31751.70MB/s
2014/04/19-13:41:45(CST) Stats: File Copy: 0.00M at 0.00MB/s
2014/04/19-13:41:45(CST) Stats: Net Copy: 0.00M at 0.00MB/s
2014/04/19-13:41:45(CST) Stats: Data Check: 0.00M at 0.00MB/s
2014/04/19-13:41:45(CST) Stats: Invert Data: 0.00M at 0.00MB/s
2014/04/19-13:41:45(CST) Stats: Disk: 0.00M at 0.00MB/s
2014/04/19-13:41:45(CST)
2014/04/19-13:41:45(CST) Status: FAIL - test discovered HW problems
2014/04/19-13:41:45(CST)
Original comment by ipmit...@gmail.com
on 21 Apr 2014 at 12:46
Can you run
stressapptest -l stressapptest.log -s 259200
That way the results will be in a log as well as on the screen. The detailed
messages have scrolled by already in the snippet you posted, that is only the
end summary.
Original comment by pud...@gmail.com
on 21 Apr 2014 at 2:04
Original issue reported on code.google.com by
ipmit...@gmail.com
on 10 Apr 2014 at 1:39Attachments: