MPSS and MICRAS service failure

[Rocha__João]

Dear all,

I'm working with Xeon Phi 7120. I tried to restore the MPSS configuration to default values and have executed the service mpss unload and cleanconfig (sudo service mpss unload and sudo micctrl --cleanconfig, respectively) but the sudo micctrl --initdefaults and sudo service mpss start don' start the mic's and mpss. I receive "exit-code" on micras.service and mpss.service.
Any help is most welcome.
Thanks.

--

João Rocha

[JJK]

most likely the mic.ko module did not load.

Please provide some info on the OS you are using and attach the output of "micdebug.sh"

[Rocha__João]

Hi JJK.
Thanks for your reply.
The OS is RedHat 3.10.0-693.el7.x86_64.
I can't run the micdebug.sh due to the MPSS config is not found. I tried to execute the micctrl --initdefaults but it gives an error as no MIC cards are found or specified.
I've attached the mpssd file from /var/log to see the errors.
Thanks.

Cheers,

João Rocha

[JJK]

there is something seriously wrong with your software stack installation. The mpssd log shows

  12025 Wed May 30 10:07:28 2018: mic1: Ignoring  - Critical failure reading state sysfs entry
  12026 Wed May 30 10:07:28 2018: mic2: Ignoring  - Critical failure reading state sysfs entry
  12027 Wed May 30 10:07:28 2018: mic3: Ignoring  - Critical failure reading state sysfs entry
  12028 Wed May 30 10:07:28 2018: mic4: Ignoring  - Critical failure reading state sysfs entry
  12029 Wed May 30 10:07:28 2018: mic5: Ignoring  - Critical failure reading state sysfs entry
  12030 Wed May 30 10:07:28 2018: mic6: Ignoring  - Critical failure reading state sysfs entry
  12031 Wed May 30 10:07:28 2018: mic7: Ignoring  - Critical failure reading state sysfs entry

that suggests that you have 8 mics present (counting start at 0) yet all of them have a "state sysfs" issue .

I'd suggest you reinstall (at least) the mpss-modules package and then reboot.

Or, better safe than sorry, update to the latest release (3.8.5)

[Rocha__João]

Hi JJK,

Thanks for your reply.
I've done the micctrl --cleanconfig because of those errors/issues.
I'll install the latest release 3.8.5 of the mpss.
Thanks a lot.
Cheers
João Rocha

[Rocha__João]

Hi JJK,

I've noticed that the 3.8.5 version of mpss is not supported for rhel 7.0 (my version of the rhel OS).
Should i try to install it anyway?

Thanks.
Best regards,
João Rocha

[JJK]

your kernel version, RedHat 3.10.0-693.el7.x86_64, is supported by mpss 3.8.5 so I'd simply try that one.

Also, if you still experience problems after reinstalling the mpss stack (most notably, mpss-modules) then try loading the mic.ko module manually (i.e.   "modprobe mic") then post the output of "dmesg"   so we can have a look at what is happening/failing.