#36 fail to run on some machine/platform

open
nobody
None
5
2005-03-23
2005-03-23
sunzen.w
No

I try to use the 2004.7.30-post version. It works on
one machine(Red Hat8.0, 3ware-9500 raid controller).
But there are problems when it is run on another
machine(RHT advance server3.0, arcmsr Raid). When
start test, a error dialog box appear and dynamo
segmentation fault.

The message of dynamo is as the following:

ot@fengshuo src]# ./dynamo -i 10.10.5.12 -m 10.10.5.117
-n test
Fail to open kstat device file

Command line parameter(s):
Looking for Iometer on "10.10.5.12"
New manager name is "test"

Sending login request...
test
10.10.5.117 (port 32779)
Login accepted.
Reporting drive information...
Physical drives (raw devices)...
Fail to get size for /dev/sda by BLKGETSIZE64
Fail to get size for /dev/sda1 by BLKGETSIZE64
Fail to get size for /dev/hda by BLKGETSIZE64
Fail to get size for /dev/hda1 by BLKGETSIZE64
Fail to get size for /dev/hda3 by BLKGETSIZE64
Fail to get size for /dev/hda4 by BLKGETSIZE64
Fail to get size for /dev/hda5 by BLKGETSIZE64
Reporting TCP network information...
done.
Worker 0 running Access Spec: 512B; 100% Read; 0% random
Access specifications for test defined.
Grunt: Growing grunt data buffer from 512 to 512
Worker 0 running Access Spec: 512B; 100% Read; 0% random
Access specifications for test defined.
Grunt: Grunt data buffer size 512 >= 512, returning
Worker 0 setting targets...
Fail to get size for /dev/sda1 by BLKGETSIZE64
Fail to get size for /dev/sda1 by BLKGETSIZE64
Fail to get size for /dev/sda1 by BLKGETSIZE64
Starting...
Segmentation fault

The main info the error dialog box is :

ERROR: Send failed.
PortTCP::SynchSend() ******** src\IOportTCP.cpp line 881
errno=10053

Discussion

  • Ming Zhang
    Ming Zhang
    2005-03-23

    Logged In: YES
    user_id=896657

    Socket error 10053 means: TCP/IP Connection Aborted by Peer

    so where is u iometer? what happens on iometer? and do u
    have firewall on it?

     
  • sunzen.w
    sunzen.w
    2005-03-24

    Logged In: YES
    user_id=912794

    It seems not because of network configuation, for iometer
    and dynamo have had communication.
    It seems Ther error just happens when dynamo start to do
    the I/O.
    Today I changed the platform from RedHaT advanced
    server3.0 to Redhat 8.0, the problem disappear and Iometer
    run well.
    Thanks.

     
  • Ming Zhang
    Ming Zhang
    2005-03-24

    Logged In: YES
    user_id=896657

    can u compile dynamo with -g flag, add a break there and
    send me the stack?

    thx

     
  • sunzen.w
    sunzen.w
    2005-03-25

    Logged In: YES
    user_id=912794

    Oh,what a pity! My platform has been changed.
    If i switch back, or the problem rehappen on some cases, I
    will do as you guide.
    Thank you.

     
  • Ming Zhang
    Ming Zhang
    2005-03-25

    Logged In: YES
    user_id=896657

    it is fine. i will try to find a rhel3 and see what i can do.

    ming

     
  • aumic
    aumic
    2005-03-30

    Logged In: YES
    user_id=1135866

    I thought it was a litter bug of iometer 2004.07.30. you should
    edit file IOCommon.h ,and change the line
    #define BLKGETSIZE64 _IOR(0x12,114, unsigned long long )
    to
    #define BLKGETSIZE64 _IOR(0x12,114, sizeof (unsigned
    long long) )

     
  • Ming Zhang
    Ming Zhang
    2005-03-30

    Logged In: YES
    user_id=896657

    see usr/include/asm/ioctl.h

    #define _IOR(type,nr,size)
    _IOC(_IOC_READ,(type),(nr),sizeof(size))

    so u can not do sizeof(sizeof())