#3436 imgcapture fails on 2.8.0

[Lissa Valletta]

Customer reported and I recreated on my 2.8 rhels6.3 machine, installed by PCM. ONly one that has a diskful installed node.
I run
[root@x3650m2n01 ~]# imgcapture dxm360m4n04
Capturing image on dxm360m4n04...
Error: The "xdsh" command fails to run "cd /; find . ! -path "./tmp/" ! -path "./proc/" ! -path "./dev/" ! -path "./sys/" ! -path "./install/" ! -path "./xcatpost/" |cpio -H newc -o |gzip -c - >/tmp/xcat_imgcapture.29554" on dxm360m4n04
[root@x3650m2n01 ~]# lsdef dxm360m4n04
Object name: dxm360m4n04
arch=x86_64
bmc=10.1.0.3
chain=runcmd=bmcsetup,osimage=rhels6.3-x86_64-stateless-compute:reboot4deploy
currchain=boot
currstate=boot
groups=__Managed,compute,__NetworkProfile_default_cn,__HardwareProfile_,__ImageProfile_rhels6.3-x86_64-stateful-compute,
initrd=xcat/osimage/rhels6.3-x86_64-stateful-compute/initrd.img
installnic=eth0
ip=10.1.0.242
kcmdline=quiet repo=http://!myipfn!:80/install/rhels6.3/x86_64 ks=http://!myipfn!:80/install/autoinst/dxm360m4n04 ksdevice=eth0
kernel=xcat/osimage/rhels6.3-x86_64-stateful-compute/vmlinuz
mac=34:40:B5:B8:E5:3A
mgt=ipmi
netboot=xnba
nichostnamesuffixes=eth0:-eth0,bmc:-bmc
nicips=bmc!10.1.0.3,eth0!10.1.0.242
nicnetworks=eth0:provision,bmc:provision
nictypes=eth0:Ethernet,bmc:BMC
os=rhels6.3
postbootscripts=syncfiles,ospkgs,otherpkgs,mountnfs,confignics
postscripts=syslog,remoteshell,syncfiles,setminiuidgid,mkresolvconf,setupntp
primarynic=eth0
profile=compute
provmethod=rhels6.3-x86_64-stateful-compute
status=booting
statustime=03-18-2013 15:28:02

This is the customers test case
imgcapture rsa03 -o SL6-x86_64-statelite-cng2 -i eth0 -n igb  -V
nodename is rsa03; os is SL6; arch is x86_64; profile is cng2
bootif is eth0; netdriver is igb
the output of "stat / -f |grep Type" on rsa03 is:
rsa03:     ID: 87cd4ce061a2fd20 Namelen: 255     Type: ext2/ext3
The file type is ext2/ext3
The excludestr is "cd /; find . ! -path "./tmp/" ! -path "./proc/" !
-path "./sys/" ! -path "./dev/" ! -path "./xcatpost/" ! -path
"./install/" |cpio -H newc -o |gzip -c - >/tmp/xcat_imgcapture.15312"
running "echo -n > /tmp/xcat_imgcapture.15312" on rsa03
Capturing image on rsa03...
running "cd /; find . ! -path "./tmp/" ! -path "./proc/" ! -path
"./sys/" ! -path "./dev/" ! -path "./xcatpost/" ! -path
"./install/" |cpio -H newc -o |gzip -c - >/tmp/xcat_imgcapture.15312"
on rsa03 via the "xdsh" command
Error: The "xdsh" command fails to run "cd /; find . ! -path "./tmp/"
! -path "./proc/" ! -path "./sys/" ! -path "./dev/" ! -path
"./xcatpost/" ! -path "./install/" |cpio -H newc -o |gzip -c -

/tmp/xcat_imgcapture.15312" on rsa03
echo $?
1

So obviously the xdsh command fails?  Let's try that directly (slight
change to use another output file):

xdsh rsa03 'cd /; find . ! -path "./tmp/" ! -path "./proc/" !
-path "./sys/" ! -path "./dev/" ! -path "./xcatpost/" ! -path
"./install/" |cpio -H newc -o |gzip -c - >/tmp/xcat_imgcapture.15313'
Error: rsa03: 4075749 blocks
echo $?

lsdef -t node rsa03
Object name: rsa03
    arch=x86_64
    bmc=rsa03-bmc
    bmcport=0
    currchain=boot
    currstate=install SL6-x86_64-cng2
    getmac=ipmi
    groups=rsa,ipmi,all
    initrd=xcat/osimage/SL6-x86_64-install-cng2/initrd.img
    installnic=eth0
    interface=eth0
    kcmdline=quiet repo=http://172.31.49.40:80/install/SL6/x86_64 ks=http://172.31.49.40:80/install/autoinst/rsa03 ksdevice=eth0 cmdline console=tty0 console=ttyS0,115200n8r
    kernel=xcat/osimage/SL6-x86_64-install-cng2/vmlinuz
    mac=6C:AE:8B:62:A0:BA
    mgt=ipmi
    mtm=7914
    netboot=xnba
    nfsserver=172.31.49.40
    os=SL6
    postbootscripts=otherpkgs,CCNIsyncfiles,CCNIconfigRepos,CCNIgpfs,CCNIgmond
    postscripts=CCNIservices,CCNIsyslog,CCNIremoteshell,CCNIautoupdate-disable,CCNIsyncfiles,CCNIibsetup,setupntp,CCNIrsa
    primarynic=eth0
    profile=cng2
    provmethod=SL6-x86_64-install-cng2
    serialflow=hard
    serialport=0
    serialspeed=115200
    status=failed
    statustime=03-13-2013 16:54:02
    tftpserver=172.31.49.40
    updatestatus=synced
    updatestatustime=03-14-2013 18:02:01

[Lissa Valletta]

There is no error but we do get data returned in the error buffer
DB<8> x $resp
0 HASH(0xa9d80a8)
'NoErrorPrefix' => 1
'error' => ARRAY(0xa9d75c8)
0 'ipv6cn1: 1530967 blocks'
'errorcode' => ARRAY(0xa9d9b80)
0 0

This causes the runxcmd_output routine to set the $::RUNCMD_RC.
Need to talk to Linda. This was a result of xdsh/xdcp returning errors in the error buffer for PCM.

[Lissa Valletta]

Decided with Linda to change the runxcmd callback routines to set $::RUNCMD_RC based on the setting of errocode in the response not the fact there is an error buffer.

[Lissa Valletta]

fixed trunk
Sending xCAT/Utils.pm
Transmitting file data .
Committed revision 15588.
fix 2.8

Sending xCAT/Utils.pm
Transmitting file data .
Committed revision 15589.

[Lissa Valletta]

The attached file can replace /opt/xcat/lib/perl/xCAT/Utils.pm in the 2.8.0 release. You should service xcatd restart after you replace it.

YOu can use the attached patch file if you prefer. cp Utils.pm.patch to the system and run
patch -p0 <Utils.pm.patch