Opened 13 years ago

Closed 12 years ago

#121 closed defect (fixed)

incorrect tmp dir used by mindi for copying gawk to

Reported by: adrianmarsh Owned by: bruno
Priority: normal Milestone: 2.2.2
Component: mindi Version: 1.2.1
Severity: normal Keywords:

Description (last modified by bruno)

Versions: (Centos 4.4)

root@ubiphone tmp.mondo.3618]# rpm -qa|egrep mindi\|mondo

During the mindi phase, it bails out with an error. Looking in mindi.log, it seems to me that mindi is trying to copy gawk to an incorrect directory (tmp )

That said, I'm running as root, so it should still be able to create /tmp.mondo.3618 (60+Gb free)

mindi.log extract:

NOT_BOOT = 'no'
Mindi's temp dir = //tmp.mondo.3618
Mindi's output dir=//mondo.scratch.19622/images
WARNING - bzip2 does not exist; cannot be LDD'd.
WARNING - lvm does not exist; cannot be LDD'd.
This warning only affects you if you are using LVM.
I don't think you are, so don't worry about it.
WARNING - lvm.static does not exist; cannot be LDD'd.


mappath = /lib/kbd/keymaps/i386/qwerty/
Added kbd map /lib/kbd/keymaps/i386/qwerty/
Added kbd map /lib/kbd/keymaps/i386/include/
Added kbd map /lib/kbd/keymaps/i386/include/
Added kbd map /lib/kbd/keymaps/mac/include/
Added kbd map /lib/kbd/keymaps/i386/include/
Added kbd map /lib/kbd/keymaps/i386/include/
Added kbd map /lib/kbd/keymaps/mac/include/
FATAL ERROR. Cannot copy ../../bin/gawk to //tmp.mondo.3618/bigdir - did you run out of disk space?

Attachments (2) (10.8 KB) - added by adrian.marsh@… 13 years ago.
mindi.log (24.8 KB) - added by adrianmarsh 13 years ago.
2nd mindi with patch

Download all attachments as: .zip

Change History (9)

Changed 13 years ago by adrian.marsh@…

comment:1 Changed 13 years ago by bruno

  • Description modified (diff)
  • Status changed from new to assigned

comment:2 Changed 13 years ago by bruno

If you're running Centos 4, could you rather try the packages available at instead ? I just want to check whether it has an impact or not.

comment:3 Changed 13 years ago by bruno

  • Milestone set to 2.2.1

Ok, I think I found it.

Could you try applying that patch to /usr/sbin/mindi:

--- mindi/mindi (révision 1019)
+++ mindi/mindi (révision 1020)
@@ -943,7 +943,7 @@
                return 0

-       link=`readlink $file`
+       link=`readlink -f $file`
        d=`dirname $file`
        if [ ! -e "$link" -a ! -e "$d/$link" ]; then
                echo "Problem with dead link on $file -> $link" >> $LOGFILE

comment:4 Changed 13 years ago by adrianmarsh

I tried the patch to the latest poste versions of mindi/mondo... Attached is the log. It doesn't look like it worked as theres still "" within the log.. but now the added problem that loop0 won't clear up.

[root@ubiphone tmp]# losetup /dev/loop0 -d ioctl: LOOP_CLR_FD: No such device or address

Changed 13 years ago by adrianmarsh

2nd mindi with patch

comment:5 Changed 13 years ago by bruno

  • Milestone changed from 2.2.1 to 2.2.2

Well he problem is completely different. 1/ The is harmless 2/ The loop0 device maybe in use by something else on your system or there may be an issue with that function and your kernel or ...

Could you start from a clean state when you have time (meaning, reboot the system and then launch the archive process). You should be able to also use the latest rhel4 version announced today which fixes the bug you're seeing.

comment:6 Changed 13 years ago by adrianmarsh

Haven't been able to try this yet... hope to later this week..

comment:7 Changed 12 years ago by bruno

  • Resolution set to fixed
  • Status changed from assigned to closed

The main part of that bug is now closed. The loop problem is now followed in #140

Note: See TracTickets for help on using tickets.