4 Replies Latest reply: May 26, 2011 7:18 AM by 520180 RSS

    VDBench Question

    520180
      Hi,

      I am trying to benchmark a StorageTech SAN using vdbench, on RHEL 5. I am just wondering if this is a valid configuration file to use for vdbench:

      sd=sd1,lun=/dev/sda1,openflags=o_direct
      wd=wd1,sd=sd*,seekpct=random, xfersize=8192,readpct=10
      rd=run1,wd=wd*,iorate=max,elapsed=30,interval=3

      The settings above are from a benchmark done on a customer site. I am trying to replicate the results on our environment. When I try and run with this configuration, vdbench crashes... I don't know how it would work anyway, since where does vdbench write to? There is no path to a file given.

      Anyway, the following does work:

      sd=sd1,lun=/raid10/quick_vdbench_test,openflags=o_direct
      wd=wd1,sd=sd1,rdpct=10,xfersize=2048
      rd=rd1,wd=wd1,iorate=max,elapsed=30,interval=3

      My questions is, should the first configuration work?

      Any help greatly appreciated.
      /John
        • 1. Re: VDBench Question
          Henk Vandenbergh-Oracle
          John,

          what do you mean with 'crashes'? If you send me the complete Vdbench output directory to vdbench@sun.com I'll take a look at it.
          Henk.
          • 2. Re: VDBench Question
            520180
            Thanks so much Henk, I have sent you a mail. The output from the java interpreter was as follows:

            stdout/stderr for slave=localhost-0

            23:01:05.770 23:01:05.769 SlaveJvm execution parameter: '-m localhost'
            23:01:05.771 23:01:05.770 SlaveJvm execution parameter: '-n localhost-10-110525-23.01.04.385'
            23:01:05.771 23:01:05.770 SlaveJvm execution parameter: '-l localhost-0'
            23:01:05.771 23:01:05.770 SlaveJvm execution parameter: '-p 5570'
            23:01:05.771 23:01:05.770 SlaveJvm positional parameter: 'SlaveJvm'
            23:01:05.812 23:01:05.812 successfully connected to master localhost
            23:01:05.812 23:01:05.812 Connection to localhost using port 5570 successful
            23:01:06.083 23:01:06.082 execute(): /home/vmdemo/tools/linux/linux_clock.suse
            23:01:06.119 23:01:06.118 ticks_per_second: 100
            23:01:06.124 23:01:06.123 Setting shared library to: /home/vmdemo/tools/linux/linux64.so
            23:01:06.124 #
            23:01:06.124 # A fatal error has been detected by the Java Runtime Environment:
            23:01:06.124 #
            23:01:06.124 # SIGSEGV (0xb) at pc=0x0000003290e79b60, pid=6371, tid=1089771840
            23:01:06.124 #
            23:01:06.124 # JRE version: 6.0_17-b17
            23:01:06.125 # Java VM: OpenJDK 64-Bit Server VM (14.0-b16 mixed mode linux-amd64 )
            23:01:06.125 # Derivative: IcedTea6 1.7.10
            23:01:06.125 # Distribution: Custom build (Thu Apr 7 12:51:17 EDT 2011)
            23:01:06.125 # Problematic frame:
            23:01:06.125 # C [libc.so.6+0x79b60] strlen+0x10
            23:01:06.125 #
            23:01:06.125 # An error report file with more information is saved as:
            23:01:06.125 # /home/vmdemo/tools/hs_err_pid6371.log
            23:01:06.128 #
            23:01:06.128 # If you would like to submit a bug report, please include
            23:01:06.128 # instructions how to reproduce the bug and visit:
            23:01:06.128 # http://icedtea.classpath.org/bugzilla
            23:01:06.128 # The crash happened outside the Java Virtual Machine in native code.
            23:01:06.129 # See problematic frame for where to report the bug.
            23:01:06.129 #
            23:01:06.131 Abort

            We are using the IcedTea jdk, don't know if that matters - have seen problems before with IcedTea. We can always install the proper java jdk if needs be.
            • 3. Re: VDBench Question
              Henk Vandenbergh-Oracle
              The opening of the raw device failed, causing Vdbench to use the strerror() function which likely returned a non-null terminated string, causing this failure. Problem is fixed in vdbench503rc11 from vdbench.org

              Henk.
              • 4. Re: VDBench Question
                520180
                Ultimately, this is a problem with writing to the device:

                chmod 777 /dev/sda

                fixes the problem

                Edited by: user517177 on May 26, 2011 5:18 AM