xCAT / Bugs / #4664 directions for install melanox IB network software fail on ppc64le

zhao er tao - 2015-06-23

assigned_to: huweihua
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

zhao er tao - 2015-06-23

Hi, Wei Hua, will you pls have a look at this bug? Thx!

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

huweihua - 2015-06-26

Hi ralph, sorry for reply later.
When I developed this IB feature for Ubuntu, I used MLNX_OFED_LINUX-2.3-1.0.1-ubuntu14.04-ppc64le.iso and Ubuntu14.04.1 on p8 server 8286-42A 101102V.
I do remember I couldn't pass the installation process without 'ARCH=powerpc'. Due to MLNX OFED is updated continuously and I couldn't test all version,
so it's possible for script mlnofed_ib_install to failed in a new environment.
so could you offer me your environment information? the version of MLNX OFED and ubuntu, the model of server. It's helpful for me to reproduce the issue.
Thank you very much.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- ralph bellofatto - 2015-06-26
  
  MLNX_OFED_LINUX-2.4-1.0.4-ubuntu14.04-ppc64le.
  
  To work around this problem we added the following line to the
  install-mellanox postscript prior to invoking the mlnxofed_ib_install
  script.
  
  This appeared to clear up the problem.
  
  Ralph Bellofatto
  IBM TJ Watson Research
  1-914-945-3321
  ralphbel@us.ibm.com
  
  From: "huweihua" huweihua@users.sf.net
  To: "[xcat:bugs] " 4664@bugs.xcat.p.re.sf.net
  Date: 06/25/2015 10:13 PM
  Subject: [xcat:bugs] #4664 directions for install melanox IB network
  software fail on ppc64le
  
  Hi ralph, sorry for reply later.
  When I developed this IB feature for Ubuntu, I used
  MLNX_OFED_LINUX-2.3-1.0.1-ubuntu14.04-ppc64le.iso and Ubuntu14.04.1 on p8
  server 8286-42A 101102V.
  I do remember I couldn't pass the installation process without
  'ARCH=powerpc'. Due to MLNX OFED is updated continuously and I couldn't
  test all version,
  so it's possible for script mlnofed_ib_install to failed in a new
  environment.
  so could you offer me your environment information? the version of MLNX
  OFED and ubuntu, the model of server. It's helpful for me to reproduce the
  issue.
  Thank you very much.
  
  [bugs:#4664] directions for install melanox IB network software fail on
  ppc64le
  
  Status: open
  Milestone: 2.9.2
  Created: Fri May 08, 2015 12:42 PM UTC by ralph bellofatto
  Last Updated: Tue Jun 23, 2015 06:32 AM UTC
  Owner: huweihua
  
  I attempted to follow the directions here for the melanox software:
  
  https://sourceforge.net/p/xcat/wiki/Managing_the_Mellanox_Infiniband_Network/
  
  However when I followed them I found the following error:
  
  tulgb004: Installing ofed-scripts-2.4...
  tulgb004: Installing mlnx-ofed-kernel-utils-2.4...
  tulgb004: Installing mlnx-ofed-kernel-dkms-2.4...
  tulgb004: Failed to install mlnx-ofed-kernel-dkms DEB
  tulgb004: Collecting debug info...
  tulgb004: See /tmp/OFED.184951.logs/mlnx-ofed-kernel-dkms.debinstall.log
  
  Further digging into the log files showed:
  
  /usr/bin/dpkg -i --force-confnew
  --force-confmiss /tmp/ofed/DEBS/mlnx-ofed-kernel-dkms_2.4-OFED.2.4.1.0.4.1.g13136df_all.deb
  
  Selecting previously unselected package mlnx-ofed-kernel-dkms.
  (Reading database ... 271277 files and directories currently installed.)
  Preparing to
  unpack .../mlnx-ofed-kernel-dkms_2.4-OFED.2.4.1.0.4.1.g13136df_all.deb ...
  Unpacking mlnx-ofed-kernel-dkms (2.4-OFED.2.4.1.0.4.1.g13136df) ...
  Setting up mlnx-ofed-kernel-dkms (2.4-OFED.2.4.1.0.4.1.g13136df) ...
  Loading new mlnx-ofed-kernel-2.4 DKMS files...
  First Installation: checking all kernels...
  Building only for 3.16.0-30-generic
  Building for architecture ppc64le
  Building initial module for 3.16.0-30-generic
  Error! Bad return status for module build on kernel: 3.16.0-30-generic
  (ppc64le)
  Consult /var/lib/dkms/mlnx-ofed-kernel/2.4/build/make.log for more
  information.
  dpkg: error processing package mlnx-ofed-kernel-dkms (--install):
  subprocess installed post-installation script returned error exit status
  10
  Errors were encountered while processing:
  mlnx-ofed-kernel-dkms
  
  We traced this back to this code
  in /opt/xcat/share/xcat/ib/scripts/Mellanox/mlnxofed_ib_install:
  
  if [ -f /etc/os-release ] && cat /etc/os-release|grep NAME|grep Ubuntu>/dev/null && uname -m |grep ppc64 >/dev/null then echo "ARCH=powerpc perl -x mlnxofedinstall --without-fw-update
  
  $mlnxofed_options"
  ARCH=powerpc perl -x mlnxofedinstall --without-fw-update
  $mlnxofed_options
  sleep 1
  service openibd restart
  else
  echo "perl -x mlnxofedinstall $mlnxofed_options"
  perl -x mlnxofedinstall $mlnxofed_options
  fi
  
  If we change the above code to:
  
  if [ -f /etc/os-release ] && cat /etc/os-release|grep NAME|grep Ubuntu>/dev/null && uname -m |grep ppc64 >/dev/null then echo "perl -x mlnxofedinstall --without-fw-update
  
  $mlnxofed_options"
  unset ARCH
  perl -x mlnxofedinstall --without-fw-update $mlnxofed_options
  sleep 1
  service openibd restart
  else
  echo "perl -x mlnxofedinstall $mlnxofed_options"
  perl -x mlnxofedinstall $mlnxofed_options
  fi
  
  The code appears to run successfully and install the drivers...
  It would appear that the dkms does NOT like the ARCH being set at all...
  
  Sent from sourceforge.net because you indicated interest in
  https://sourceforge.net/p/xcat/bugs/4664/
  
  To unsubscribe from further messages, please visit
  https://sourceforge.net/auth/subscriptions/
  
  Related
  
  Bugs: #4664
  
  alternate
  
  graycol.gif
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- ralph bellofatto - 2015-06-26
  
  we also patched this into the mellanox_ib_install script.
  
  #RB... fix this, #echo "ARCH=powerpc perl -x mlnxofedinstall --without-fw-update
  
  $mlnxofed_options"
  #ARCH=powerpc perl -x mlnxofedinstall --without-fw-update
  $mlnxofed_options
  echo "perl -x mlnxofedinstall --without-fw-update
  $mlnxofed_options"
  unset ARCH
  rc=0
  perl -x mlnxofedinstall --without-fw-update $mlnxofed_options
  || rc=$?
  
  Ralph Bellofatto
  IBM TJ Watson Research
  1-914-945-3321
  ralphbel@us.ibm.com
  
  From: "huweihua" huweihua@users.sf.net
  To: "[xcat:bugs] " 4664@bugs.xcat.p.re.sf.net
  Date: 06/25/2015 10:13 PM
  Subject: [xcat:bugs] #4664 directions for install melanox IB network
  software fail on ppc64le
  
  Hi ralph, sorry for reply later.
  When I developed this IB feature for Ubuntu, I used
  MLNX_OFED_LINUX-2.3-1.0.1-ubuntu14.04-ppc64le.iso and Ubuntu14.04.1 on p8
  server 8286-42A 101102V.
  I do remember I couldn't pass the installation process without
  'ARCH=powerpc'. Due to MLNX OFED is updated continuously and I couldn't
  test all version,
  so it's possible for script mlnofed_ib_install to failed in a new
  environment.
  so could you offer me your environment information? the version of MLNX
  OFED and ubuntu, the model of server. It's helpful for me to reproduce the
  issue.
  Thank you very much.
  
  [bugs:#4664] directions for install melanox IB network software fail on
  ppc64le
  
  Status: open
  Milestone: 2.9.2
  Created: Fri May 08, 2015 12:42 PM UTC by ralph bellofatto
  Last Updated: Tue Jun 23, 2015 06:32 AM UTC
  Owner: huweihua
  
  I attempted to follow the directions here for the melanox software:
  
  https://sourceforge.net/p/xcat/wiki/Managing_the_Mellanox_Infiniband_Network/
  
  However when I followed them I found the following error:
  
  tulgb004: Installing ofed-scripts-2.4...
  tulgb004: Installing mlnx-ofed-kernel-utils-2.4...
  tulgb004: Installing mlnx-ofed-kernel-dkms-2.4...
  tulgb004: Failed to install mlnx-ofed-kernel-dkms DEB
  tulgb004: Collecting debug info...
  tulgb004: See /tmp/OFED.184951.logs/mlnx-ofed-kernel-dkms.debinstall.log
  
  Further digging into the log files showed:
  
  /usr/bin/dpkg -i --force-confnew
  --force-confmiss /tmp/ofed/DEBS/mlnx-ofed-kernel-dkms_2.4-OFED.2.4.1.0.4.1.g13136df_all.deb
  
  Selecting previously unselected package mlnx-ofed-kernel-dkms.
  (Reading database ... 271277 files and directories currently installed.)
  Preparing to
  unpack .../mlnx-ofed-kernel-dkms_2.4-OFED.2.4.1.0.4.1.g13136df_all.deb ...
  Unpacking mlnx-ofed-kernel-dkms (2.4-OFED.2.4.1.0.4.1.g13136df) ...
  Setting up mlnx-ofed-kernel-dkms (2.4-OFED.2.4.1.0.4.1.g13136df) ...
  Loading new mlnx-ofed-kernel-2.4 DKMS files...
  First Installation: checking all kernels...
  Building only for 3.16.0-30-generic
  Building for architecture ppc64le
  Building initial module for 3.16.0-30-generic
  Error! Bad return status for module build on kernel: 3.16.0-30-generic
  (ppc64le)
  Consult /var/lib/dkms/mlnx-ofed-kernel/2.4/build/make.log for more
  information.
  dpkg: error processing package mlnx-ofed-kernel-dkms (--install):
  subprocess installed post-installation script returned error exit status
  10
  Errors were encountered while processing:
  mlnx-ofed-kernel-dkms
  
  We traced this back to this code
  in /opt/xcat/share/xcat/ib/scripts/Mellanox/mlnxofed_ib_install:
  
  if [ -f /etc/os-release ] && cat /etc/os-release|grep NAME|grep Ubuntu>/dev/null && uname -m |grep ppc64 >/dev/null then echo "ARCH=powerpc perl -x mlnxofedinstall --without-fw-update
  
  $mlnxofed_options"
  ARCH=powerpc perl -x mlnxofedinstall --without-fw-update
  $mlnxofed_options
  sleep 1
  service openibd restart
  else
  echo "perl -x mlnxofedinstall $mlnxofed_options"
  perl -x mlnxofedinstall $mlnxofed_options
  fi
  
  If we change the above code to:
  
  if [ -f /etc/os-release ] && cat /etc/os-release|grep NAME|grep Ubuntu>/dev/null && uname -m |grep ppc64 >/dev/null then echo "perl -x mlnxofedinstall --without-fw-update
  
  $mlnxofed_options"
  unset ARCH
  perl -x mlnxofedinstall --without-fw-update $mlnxofed_options
  sleep 1
  service openibd restart
  else
  echo "perl -x mlnxofedinstall $mlnxofed_options"
  perl -x mlnxofedinstall $mlnxofed_options
  fi
  
  The code appears to run successfully and install the drivers...
  It would appear that the dkms does NOT like the ARCH being set at all...
  
  Sent from sourceforge.net because you indicated interest in
  https://sourceforge.net/p/xcat/bugs/4664/
  
  To unsubscribe from further messages, please visit
  https://sourceforge.net/auth/subscriptions/
  
  Related
  
  Bugs: #4664
  
  alternate
  
  graycol.gif
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

directions for install melanox IB network software fail on ppc64le

An extreme cluster/cloud administration toolkit

Milestone

Searches

Help

#4664 directions for install melanox IB network software fail on ppc64le

Related

Discussion

Related

Related