Can't get ALL togeather

  • Andrei Gololobov

    I USE :
    pylesos tools # uname -a
    Linux pylesos 2.6.25-gentoo-r6 #1 SMP Mon Oct 20 16:25:03 EEST 2008 x86_64 AMD Sempron(tm) Processor 2600+ AuthenticAMD GNU/Linux
    pylesos tools # java-config -L
    The following VMs are available for generation-2:
    *) Sun JDK [sun-jdk-1.6]
    pylesos tools # emerge -pv boost log4cpp cmake gcc ant sys-fs/fuse
    These are the packages that would be merged, in order:
    Calculating dependencies... done!
    [ebuild U ] sys-devel/gcc-config-1.4.0-r4 [1.3.16] 0 kB 
    [ebuild R ] dev-libs/boost-1.34.1-r2 USE="-debug -doc -icu -pyste -tools" 0 kB 
    [ebuild R ] dev-libs/log4cpp-1.0 USE="-doc" 0 kB 
    [ebuild R ] dev-util/cmake-2.4.6-r1 USE="-emacs -vim-syntax" 0 kB 
    [ebuild R ] dev-java/ant-1.7.0 0 kB 
    [ebuild UD] sys-fs/fuse-2.7.0 [2.7.3] 491 kB 
    [ebuild R ] sys-devel/gcc-4.1.2 USE="fortran gtk mudflap nls (-altivec) -bootstrap -build -d -doc -gcj (-hardened) -ip28 -ip32r10k -libffi (-multilib) -multislot (-n32) (-n64) -nocxx -objc -objc++ -objc-gc -test -vanilla" 50 kB 

    I have 2 nodes installed and compiled, But the path to kfs differs:
    pylesos scripts # cat ./machines.cfg 
    # KFS Machine configuration file
    node: pylesos
    rundir: /home/kfs/kfs-0.2.2/meta 
    baseport: 20000
    backup_path: pylesos:/home/kfs/kfs-0.2.2/meta_bkup
    clusterkey: kfs-pylesos-cluster
    loglevel: DEBUG
    numservers: 2
    node: pylesos
    rundir: /home/kfs/kfs-0.2.2/chunk
    chunkDir: /home/kfs/kfs-0.2.2/chunk/kfs1 /home/kfs/kfs-0.2.2/chunk/kfs2 
    baseport: 30000 
    loglevel: DEBUG
    space: 20 G

    node: barbus
    rundir: /home/ada/kfs-0.2.2/chunk
    chunkDir: /home/ada/kfs-0.2.2/chunk/kfs1
    baseport: 30000 
    loglevel: DEBUG
    space: 2 G

    the configs on both servers are similar.
    But when I try to start... here is what I get:
    pylesos scripts # python -f machines.cfg -s
    Running cmd ssh barbus 'cd /home/ada/kfs-0.2.2/chunk; scripts/ -s -c -f bin/ChunkServer.prp ' 
    Running cmd ssh pylesos 'cd /home/kfs/kfs-0.2.2/meta; scripts/ -s -m -f bin/MetaServer.prp ' 
    Running cmd ssh pylesos 'cd /home/kfs/kfs-0.2.2/chunk; scripts/ -s -c -f bin/ChunkServer.prp ' 
    Started all the workers..waiting for them to finish
    bash: scripts/ No such file or directory
    pylesos: Starting chunkserver...
    cleaner is already running...

    pylesos: Starting metaserver...
    pylesos: Starting cleaner...

    The one of the scripts are not found and when we check we get:
    pylesos tools # ./kfsping -m -s pylesos -p 20000
    Up servers: 1
    s= , p=30000, total=20(GB), used=0(GB), util=0%, nblocks=0, lastheard=8 (sec), ncorrupt=0, nchunksToMove=0
    It means that only 1 of 2 nodes start. 


    Second of all I don't get the webui started. The conf files in the webui are ok , but the tmp.txt in webui folder says:
    pylesos webui # cat ./tmp.txt 
    File "webui/", line 42
    class ServerLocation():
    SyntaxError: invalid syntax
    2) WHY DOESN'T WEBUI Start?

    The last that doesn't work is KFS_FUSE

    ylesos bin # cat ./kfs.prp = pylesos
    metaServer.port = 20000

    pylesos bin # ./kfs_fuse /mnt/kfs/ -d
    unique: 1, opcode: INIT (26), nodeid: 0, insize: 56
    INIT: 7.9
    10-21-2008 18:15:56.919 ERROR - (/home/kfs/kfs-0.2.2/src/cc/libkfsClient/ Unable to connect to metaserver at: :20000 
    kfs_fuse: /usr/include/boost/shared_ptr.hpp:315: T* boost::shared_ptr<T>::operator->() const [with T = KFS::KfsClient]: Assertion `px != 0' failed.

    Guys I NEED HELP, can anyone explain what I do wrong?

    • sriramsrao

      sriramsrao - 2009-03-01

      For #1: The paths don't have to be the same.

      For #2: the fix has been checked into 0.2.3

      For #3: your kfs.prp is incorrect.  It should be: = pylesos
      metaServer.port = 20000


Log in to post a comment.