Bccd-ng - January-28-2012

From BCCD 3.0

Jump to: navigation, search

Contents

BCCD ISO TESTING REPORT

Author Mobeen Ludin
Date Generated January 28 / 2012
Date Tested January 28 / 2012
Release Revision # r3553

Summary

--This iso was generated to fix some of the network issues that new BCCD users were having.

--Skylar made the changes with regards to Karl, and Brad's comments and suggestions on the bug.

--This fix problems with rpc.statd starting before the loopback interface is online in diskless in mode.

--In this ISO (only 64-bit) MPI is working, Bluetooth is working ok.


Downloading and Making Bootable USB

First Downloading the ISO form a repository on earlham cs cluster:

first downloading the script for making bootable USB drive

  $ svn co svn+ssh://mmludin08@cluster.earlham.edu/cluster/svnroot/bccd-ng/trunk/bin BCCD
  $ cd BCCD 

This is where the the script exits for making bootable USB from BCCD ISO

Download the ISO from appropriate links

  $ sudo wget http://cluster.earlham.edu/bccd-ng/snapshot/bccd_latest.amd64.iso.md5
  $ sudo wget http://cluster.earlham.edu/bccd-ng/snapshot/bccd_latest.amd64.iso

CheckSum Report:

To see if the ISO was downloaded 100% correctly

  $ md5sum bccd_latest.amd64.iso

de48ef7ca8e3aff3a6474b3c9262bdf8 bccd_latest.amd64.iso

  $ cat bccd_latest.amd64.iso.md5 

MD5 (bccd_latest.amd64.iso) = de48ef7ca8e3aff3a6474b3c9262bdf8

NOTE:if can compare by reading: copy both results to two seperate files.

And run diff on them. It will tell you the difference if exist.

Second Making Bootable USB

The next step will be to insert the 4G USB drive (the Intel one) into the machine, wait a couple seconds, and then do

   $ sudo ./build_bootable_USB.sh /dev/sdb bccd_latest.amd64.iso

This is the usual procedure for running the script. The thing that can vary is /dev/sdb; usually this will be the location of the USB, but if there is another USB connected it might actually be /dev/sdc.

NOTE: To see find out the memory partitions run the following command:

   $ ls /dev/sd*

or

    $ dmesg

ISO TESTING REPORT

LIVEMODE --From USB

64-bit

      Test_suite: Failed. 
       GalaxSee, Life, Param_space, and CUDA programs worked fine

Network Managment

     $ bccd-nic-setup 

works fine

$ sudo /bin/bccd-reset-network

Worked fine: and generated no errors.

at /bin/bccd-nic-setup line 65

ok 11 - Starting networking

Configuring network interfaces...done.

Configuring network interfaces...done.

Starting ISC DHCP server: dhcpd.

Starting HPA's tftpd: in.tftpd.

Liberating BCCD

$ sudo perl /root/liberate.pl --libdev /dev/sda

Liberation Error

Liberation failed on test 8. same way as reported before.

[Got this error: ERROR START]

not ok 8 - Set LVM PV partition type

WARNING: DOS-compatible mode is deprecated. It's strongly recommended to switch off the mode (command 'c') and change display units to sectors (command 'u').

Command (m for help): Partition number (1-4): Hex code (type L to list codes): Changed system type of partition 2 to 8e (Linux LVM)

Command (m for help): The partition table has been altered!

Calling ioctl() to re-read partition table.

Warning: Re-reading the partition table failed with error 16: Device or resource busy.

The kernel still uses the old table. The new table will be used at the next reboot or after you run partprobe(8) or kpartx(8) Syncing disks.

...

Passed 77 out of 78 total tests.

[ERROR END ]


LIBERATED -- From Hard Disk

64-bit

Network Setup Managment

$ bccd-nic-setup $ sudo /bin/bccd-reset-network

Network setup workd smoothly. No Errors.

PXE-BOOTING

   - PXE-Booting is working without any problem. On my LittleFe, all the six nodes are up and GalaxSee is running on them. 
   
   - I will work on the Test_suite tomorrow (Dec 20/2011). 
   
   - openmpi, mpich2 modules worked fine. 

Errors

When i compiled the Cuda module, I got the error for symbolic link missing:

Here is the compiler error:

  /usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/4.4.5/libstdc++.so when searching for -lstdc++

Here is the soft-link that fixes the problem:

     $ sudo ln -s /usr/lib32/libstdc++.so.6 /usr/lib32/libstdc++.so

Usage

Somehow when I booted a diskless I got prompted for the usage survey. This is despite /etc/init.d/bccd-usage having code that should only run the script on node000 and in LIVE mode.

Final: Comments/Suggestions/Objection/Question/?/?

  -  Bluetoot is still picking up all the bluetood devices around. 
  -  For the New GalaxSee (GALAXSEEHPC) we need these packages to come with BCCD:

- We need to add:

* latest atlas: atlas3.9.51

* latest fftw: fftw-3.3

            *  opengl libraries
            *
Personal tools
Namespaces
Variants
Actions
Navigation
Toolbox