mondo-devel Mailing List for Mondo Rescue

FYI...

There are no packages under /test/suse11.1

Feature or bug?

Jerry Chapman said on Fri, Aug 28, 2009 at 07:31:34PM -0400:

> ftp site.  The package manager keeps wanting to upgrade (actually downgrade)
> mindi and mondo every time, I wish I could stop this behavior.

Cf:
http://trac.mondorescue.org/wiki/FAQ#Q11DoesmondoworkwithDebianUbuntudistributions

> It would be nice if those latest packages for mondo and mindi were present
> in the 9.04 directory.  I had to go looking for them.

Yes. But my Ubuntu 9.04 VMs do not work at the moment (grub issues).
Will have to llok t it for 2.2.9 generation, so shouldn't be too long
now.

> Keep up the good work, guys.

Thanks !

Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Hi,

I just wanted to say that the problems disappeared when i used the packages
i found in ftp:\\ftp.mondorescue.org\ubuntu\8.10.  I installed the latest
mindi and mondo I found there and everything worked for me.  Basically, I
did an apt-get install mondo which brought over all the libraries, then I
manually installed mindi and mondo again from the packages I found on the
ftp site.  The package manager keeps wanting to upgrade (actually downgrade)
mindi and mondo every time, I wish I could stop this behavior.

It would be nice if those latest packages for mondo and mindi were present
in the 9.04 directory.  I had to go looking for them.

By the way, I'm actually running Linux Mint 7 with the Karmic repositories
of Ubuntu, so there are a lot of alpha packages enabled.

Keep up the good work, guys.

On Mon, Aug 24, 2009 at 8:50 PM, Bruno Cornec <Bru...@hp...> wrote:

> Jerry Chapman said on Mon, Aug 24, 2009 at 08:49:46AM -0400:
>
> > I get dropped to a busybox shell when making a boot disk from
> mondoarchive
> > or from mindi itself.  I have attached mondorestore.gz and messages.gz.
>  I
>
> I'd also need the mondoarchive.log please.
>
> > cannot seem to mount my hard disk at the shell, but I managed to mount a
> usb
> > drive to copy over the logs.
>
> Yep. No HDD is seen.
>
> Any reason to not use mindi 2.0.6 ?
> What is your distro ?
> Bruno.
> --
> Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol.
> Center
> http://www.mondorescue.org / HP/Intel Solution Center \
> http://hpintelco.net
> Des infos sur Linux?  http://www.HyPer-Linux.org
> http://www.hp.com/linux
> La musique ancienne?  http://www.musique-ancienne.org
> http://www.medieval.org
>
>
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
> trial. Simplify your report design, integration and deployment - and focus
> on
> what you do best, core application coding. Discover what's new with
> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> _______________________________________________
> Mondo-devel mailing list
> Mon...@li...
> https://lists.sourceforge.net/lists/listinfo/mondo-devel
>

Andy said on Wed, Aug 26, 2009 at 02:47:24PM +0100:

> I have a system (CentOS 4.6 i386) where /var/log is a link to /data/log, where /data is in a separate partition.  I'm using latest
> mondo rescue tool set.  The mondoarchive appears to run OK, but restoring to a clean disk from the iso does not go too well.  Again,
> it seems to run through OK, but when I reboot, I start getting weird errors.  This is because in the restored system, /var/log is no
> longer a link to /data/log, but is a text file containing this:

I see. Indeed I never tested such a conf, so there may well be a bug
somewhere for that type of setup.

Of course, on a side note, why not just mount the partition directly
under /var/log, instead of doing a link ?

> I have (I hope!) worked around this by:
> cd /var ; mv log mondorestore.log ; ln -s /data/log /var/log

Yes that's the right think to do.

> and rebooting, but this looks like a mondo bug to me.

Yes. You can enter it on trac. And what would be really useful is your
full mondorestore.log file attached here or to the BR.

TIA,
Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Chet Rebman said on Thu, Aug 27, 2009 at 11:30:21AM -0600:

> Bruno, I am now using Mindi v2.0.6-r2153, which when booting gives the error...
>   Kernel panic - not syncing: No init found. Try passing init = option to kernel

I don't really understand your env.
You use mindi 2.0.6 with mondo 2.2.7 with mindi-busybox 1.10.x

Could you try to remove all those packages and just install the packages 
available at ftp://ftp.mondorescue.org/ubuntu please.

Also your log mention /usr/local paths, which shouldn't be the case if
you use packages. So I don't know why it's like that.

> Mindi-BusyBox v1.10.2 (Ubuntu 1:1.10.2-2ubuntu7) multi-call binary

This one is clearly not supported by me. Maybe you should also ask on
Ubuntu forums where the people who provide hose packages may help.

> MONDO_SHARE = /usr/share/mondo

Ok.

> MINDI_LIB = /usr/local/lib/mindi
> MINDI_SBIN = /usr/local/sbin
> MINDI_CONF = /usr/local/etc/mindi

Strange.

> lsmod result:
> -------------

I've not found your HDD module. Which one is it ?

> Module                  Size  Used by
> heci                   65680  0 

Is it this one ?

Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Bruno, I am now using Mindi v2.0.6-r2153, which when booting gives the error...

  Kernel panic - not syncing: No init found. Try passing init = option to kernel

After receiving this error, I tried to change the busybox and monoarchive versions to match and accidently 
removed some libraries, fortunately I was able to use the non-bootable backups to restore the system, however 
a boot disk would really be nice :-)

I have attached mondoarchive.log

Thanks, Chet

________________________________________
From: Bruno Cornec [Bru...@hp...]
Sent: Monday, August 24, 2009 5:35 PM
To: Mondo mailing list
Subject: Re: [Mondo-devel] mindi cd boot failure

Hello,

Chet Rebman said on Mon, Aug 24, 2009 at 12:48:44PM -0600:
> **** my mondoarchive back scripts work great, I am able to browse the ISO's and restore files
> using mondorestore. However I am unable to boot from the CDs in order to perform a compare or a
> bare-metal restore. The boot gives the following error...
>
>       sh: can't access tty; job control turned off

So your mondo CD is not found.

> **** Since it is a Debian like systems I tried the switch -k FAILSAFE and get the following error
> upon creation...

No. Debian used to need that, but not anymore.

>       Mindi failed to create your boot+data disks.
>       Fatal error... Failed to generate boot+data disks
>       ---FATALERROR--- Failed to generate boot+data disks

I'd need your mondoarchive.log file as well.

>       Mindi v2.0.4-r2045

Could you use mindi 2.0.6 for Ubuntu

>       Ubuntu 9.04

Sorry, no VM at the moment for that version, but previous version will
work (Cf: ftp://ftp.mondorescue.org/ubuntu)

Bruno.
--
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Mondo-devel mailing list
Mon...@li...
https://lists.sourceforge.net/lists/listinfo/mondo-devel

I have a system (CentOS 4.6 i386) where /var/log is a link to /data/log, where /data is in a separate partition.  I'm using latest
mondo rescue tool set.  The mondoarchive appears to run OK, but restoring to a clean disk from the iso does not go too well.  Again,
it seems to run through OK, but when I reboot, I start getting weird errors.  This is because in the restored system, /var/log is no
longer a link to /data/log, but is a text file containing this:
 
12:03 Extracted additional /dev entries OK.
12:03 Trying 'mount /dev/shm -t tmpfs -o size=64m'
12:03 Great. Pivot succeeded w/ size=64m
12:03 Successfully mounted dynamic /tmp ramdisk
12:04 find-and-mount-cdrom --- starting
12:04 Trying /dev/hda
12:04 It's not in /dev/hda; I'll keep looking
12:04 Trying /dev/hdb
12:04 It's not in /dev/hdb; I'll keep looking
12:04 Trying /dev/hdc
12:04 /dev/hdc has a CD-ROM in it
12:04 CD-ROM found at /dev/hdc
12:04 /dev/hdc is where the Mondo CD lives.
12:04 CD-ROM found and mounted at /dev/hdc
12:04 find-and-mount-cdrom --- leaving (0)
12:04 OK, I am running on a CD-ROM. Good.
12:04 Starting install-additional-tools
12:04 PLEASE WAIT. Untarring data content
12:04 Populating / with tar file content...
12:04 Done.
etc etc
 
Basically, this looks like the contents of the mondo restore log.
 
Now anything else on my system trying to write to /var/log/... isn't able to.
 
I have (I hope!) worked around this by:

cd /var ; mv log mondorestore.log ; ln -s /data/log /var/log

and rebooting, but this looks like a mondo bug to me.

What does anyone else think ?

thanks, Andy

Hello Matt !

Matt Anderson said on Tue, Aug 25, 2009 at 11:11:04AM -0400:

> >     This is one of the tricky things about selinux. An admin can
> >     redirect output from a confined domain to any directory, So writing
> >     policy to allow output to all possible file_types is not good
> >     security or policy. So this problem is really a difficult problem to
> >     solve. Allow confined domains to write to /tmp just for redirection
> >     might not seem unreasonable, but this could be an attack vector from
> >     a confined domains against users.
> 
> One possible way around this problem would be to incorporate the policy 
> changes with an SELinux boolean.  This way the default setting for the 
> policy can remain restrictive, but an Admin can temporarily loosen the 
> restrictions by adjusting the boolean allowing the backup to occur. 
> Once the backup is done the boolean could be reset back, limiting the 
> exposure on the system.

Happy to get your feedback here !
That would be indeed a good idea. Of course t's something we need to
push upstream I think ?

I personaly don't see why it would be more secure to unallow lvm command
output to a /tmp/xxx.xxx dir and still allow to pipe it to cat to the
same dir :-)

Let me know how you think we can work on solving that issue.

Best regards,
Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

> Hum, not sure about it. I found a discussion on this "Targeted policy
> does not allow lvm output redirection ?"
> (http://markmail.org/message/iqcj7jlera5vjtoz); where they said this:
>
>         Should I open a Bugzilla for this ?
>
>     This is one of the tricky things about selinux. An admin can
>     redirect output from a confined domain to any directory, So writing
>     policy to allow output to all possible file_types is not good
>     security or policy. So this problem is really a difficult problem to
>     solve. Allow confined domains to write to /tmp just for redirection
>     might not seem unreasonable, but this could be an attack vector from
>     a confined domains against users.

One possible way around this problem would be to incorporate the policy 
changes with an SELinux boolean.  This way the default setting for the 
policy can remain restrictive, but an Admin can temporarily loosen the 
restrictions by adjusting the boolean allowing the backup to occur. 
Once the backup is done the boolean could be reset back, limiting the 
exposure on the system.

-matt

Jerry Chapman said on Mon, Aug 24, 2009 at 08:49:46AM -0400:

> I get dropped to a busybox shell when making a boot disk from mondoarchive
> or from mindi itself.  I have attached mondorestore.gz and messages.gz.  I

I'd also need the mondoarchive.log please.

> cannot seem to mount my hard disk at the shell, but I managed to mount a usb
> drive to copy over the logs.

Yep. No HDD is seen.

Any reason to not use mindi 2.0.6 ?
What is your distro ?
Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Hello,

Chet Rebman said on Mon, Aug 24, 2009 at 12:48:44PM -0600:
> **** my mondoarchive back scripts work great, I am able to browse the ISO's and restore files
> using mondorestore. However I am unable to boot from the CDs in order to perform a compare or a 
> bare-metal restore. The boot gives the following error...
> 
> 	sh: can't access tty; job control turned off

So your mondo CD is not found.

> **** Since it is a Debian like systems I tried the switch -k FAILSAFE and get the following error
> upon creation...

No. Debian used to need that, but not anymore.

> 	Mindi failed to create your boot+data disks.
> 	Fatal error... Failed to generate boot+data disks
> 	---FATALERROR--- Failed to generate boot+data disks

I'd need your mondoarchive.log file as well.

> 	Mindi v2.0.4-r2045

Could you use mindi 2.0.6 for Ubuntu

> 	Ubuntu 9.04

Sorry, no VM at the moment for that version, but previous version will
work (Cf: ftp://ftp.mondorescue.org/ubuntu)

Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

I have found references to this problem on the web, however I have been unable to resolve it.

**** my mondoarchive back scripts work great, I am able to browse the ISO's and restore files
using mondorestore. However I am unable to boot from the CDs in order to perform a compare or a 
bare-metal restore. The boot gives the following error...

	sh: can't access tty; job control turned off

**** Since it is a Debian like systems I tried the switch -k FAILSAFE and get the following error
upon creation...

	Calling MINDI to create boot+data disks
	Your boot loader is GRUB and it boots from /dev/sda
	Waiting for external binary to start
	Mindi failed to create your boot+data disks.
	Fatal error... Failed to generate boot+data disks
	---FATALERROR--- Failed to generate boot+data disks

****

	Mindi v2.0.4-r2045

	mondoarchive v2.2.7-r2045

	Ubuntu 9.04
                
	kernel 2.6.28-15-generic

thanks..chet

Hi,

Some informations could help the community to help you:
* o.s. and version, Mondo and mindi versions, mondo logs...
* do you restore on the same machine ?

During mondorestore, it should be possible for you to switch to a busybox shell, through Alt-F2, in order to check things (check /var/log/mondorestore.log, and save it on a mounted USB key, or tftp it to a tftp server).

The iso files created by mondoarchive can be opened and browsed by 7-zip (http://www.7-zip.org).

Regards,
Victor

-----Original Message-----
From: Dean Yorke [mailto:dea...@xy...] 
Sent: samedi 22 août 2009 07:29
To: mon...@li...
Subject: [Mondo-devel] problem

HI,

I can't get the backup to work.

It keeps saying that it can't mount my sda1 and sda2

what do I need to do to fix this.

Is there a way to restore the files manually?



------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Mondo-devel mailing list
Mon...@li...
https://lists.sourceforge.net/lists/listinfo/mondo-devel

Dean Yorke said on Sat, Aug 22, 2009 at 01:29:12AM -0400:
> what do I need to do to fix this.

Read. Especially: http://www.mondorescue.org/support.shtml

Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Pablo said on Sat, Aug 22, 2009 at 12:55:10PM +0200:
> Just allowed my system (Ubuntu) to update itself. Does this mean I have 
> to go get the "older" (newer) version from the Mondo site? Or has the 
> version label thingy been fixed?

Cf:
http://trac.mondorescue.org/wiki/FAQ#Q11DoesmondoworkwithDebianUbuntudistributions

Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Not really my area (I am running Mandriva) but I would say No.  The problem 
was something like this: version 2.2.4 was put out as 2.24.  If that was not 
the version you had installed immediately before the upgrade, there should be 
no problem.  The Mondo Web site says that all you needed to do is "apt-get 
install mondo." http://www.mondorescue.org/downloads.shtml

HTH,

Doug.
On Sat, 22 Aug 2009 08:55:10 pm Pablo wrote:
> Just allowed my system (Ubuntu) to update itself. Does this mean I have
> to go get the "older" (newer) version from the Mondo site? Or has the
> version label thingy been fixed?
>
> Cheers.
>
> Paul.
>
> ---------------------------------------------------------------------------
>--- Let Crystal Reports handle the reporting - Free Crystal Reports 2008
> 30-Day trial. Simplify your report design, integration and deployment - and
> focus on what you do best, core application coding. Discover what's new
> with Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> _______________________________________________
> Mondo-devel mailing list
> Mon...@li...
> https://lists.sourceforge.net/lists/listinfo/mondo-devel
>
> !DSPAM:4a8ff00a211792112899914!

Just allowed my system (Ubuntu) to update itself. Does this mean I have 
to go get the "older" (newer) version from the Mondo site? Or has the 
version label thingy been fixed?

Cheers.

Paul.

HI,

I can't get the backup to work.

It keeps saying that it can't mount my sda1 and sda2

what do I need to do to fix this.

Is there a way to restore the files manually?

Bruno,

2009/8/21 Bruno Cornec <Bru...@hp...>

>
> Hummm. Should we open a Bug Report for that for Fedora/RHEL ?
>
> > >     altogether. Disabling SELinux protection is not recommended. Please
> file a
> > >     http://bugzilla.redhat.com/bugzilla/enter_bug.cgi against this
> package.
>
> Maybe I should no ?


Hum, not sure about it. I found a discussion on this "Targeted policy does
not allow lvm output redirection ?" (
http://markmail.org/message/iqcj7jlera5vjtoz); where they said this:

Should I open a Bugzilla for this ?
>>
> This is one of the tricky things about selinux. An admin can redirect
> output from a confined domain to any directory, So writing policy to allow
> output to all possible file_types is not good security or policy. So this
> problem is really a difficult problem to solve. Allow confined domains to
> write to /tmp just for redirection might not seem unreasonable, but this
> could be an attack vector from a confined domains against users.
>
Bàt
Laurent

On 21 August 2009 at 16:47, Bruno Cornec <Bru...@hp...> wrote:

> > Bruno, hope you had a great vacation!
> 
> Yes thanks ! Even mondo benefited from it.

I hope you and your family benefited the most, not that it's a
contest, but real relaxation can be very recharging.

> I wrote a small script to measure the quality of the code, the
> way I see it, and improvements I plan to do.

What you wrote seems like very good metrics, nice job!  I wonder
if any other metrics are useful and easy to collect, measure and
improve upon?  That last comment is directed at the community,
hoping that some might have experience in software quality
management to draw from.

Thanks....

--
Kevin

Miles Tillinger said on Fri, Aug 21, 2009 at 04:44:32PM +1000:

> Would you mind posting how I can search for these filenames that are causing
> the EOF error?  I am having the same problem but not sure how to find and
> remove the offending files so my backups work until 2.2.10 is released with
> the fix...

Try looking at /var/log/mondoarchive.log file for opendir errors.

grep -E '^opendir ' /var/log/mondoarchive.log

That should point you to the culprits.

Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Hugo Vanwoerkom said on Fri, Aug 21, 2009 at 09:21:48AM -0500:
> Bruno, hope you had a great vacation!

Yes thanks ! Even mondo benefited from it. I wrote a small script to
measure the quality of the code, the way I see it, and improvements I
plan to do. Here are the results:

victoria2.home.musique-ancienne.org-bruno ~/mondo/svn/branches/2.2.8 > tools/quality | grep total
total asprintf usage : 149
total mr_asprintf usage : 0
total sprintf usage : 998
total snprintf usage : 9
total strcat usage : 133
total strcpy usage : 682
total strncpy usage : 30
total fgets usage : 83
total malloc usage : 154
total mr_malloc usage : 0
total malloc_string usage : 519
total getline usage : 5
total mr_getline usage : 0
total MAX_STR_LEN usage : 430
total getcwd usage : 14
total goto usage : 31
total free usage : 42
total mr_free usage : 0
total paranoid_free usage : 833
total paranoid_system usage : 69
total mr_system usage : 0
total assert usage : 299

I introduced dynamic memory allocation usage for 2.2.9, instead of
everything static as in 2.2.8. Of course, it may create more easily core
dumps, if pointers are not initialized, but memory consumed is allocate
donly on the fly (so better), and buffer overflows are eliminated (so
much better). So between Feb 2009 and now here is what has been done:

victoria2.home.musique-ancienne.org-bruno ~/mondo/svn/branches/2.2.9 > tools/quality | grep total
total asprintf usage : 2
total mr_asprintf usage : 470       /// our version of asprintf
total sprintf usage : 813           /// still work to do
total snprintf usage : 11
total strcat usage : 33             /// a bit less dangerous
total strcpy usage : 551            ///  but still here !
total strncpy usage : 32
total fgets usage : 83
total malloc usage : 151
total mr_malloc usage : 10
total malloc_string usage : 407     /// target is to remove it completely 
total getline usage : 7
total mr_getline usage : 3
total MAX_STR_LEN usage : 421       /// same here
total getcwd usage : 14
total goto usage : 27
total free usage : 40
total mr_free usage : 200
total paranoid_free usage : 806
total paranoid_system usage : 67
total mr_system usage : 0
total assert usage : 292

That branch is now frosen. I'm still waiting fedbacks on 2 bug reports
to be sure it can be pushed as stable. Feel free to test the current
beta at ftp://ftp.mondorescue.org/test ! I wish to publish it ASAP.

Now, if you look at what I did during my vacation (outside from going to
the beach, playing with kids, preparing some lunches, visiting
monuments, ... :-)

victoria2.home.musique-ancienne.org-bruno ~/mondo/svn/branches/2.2.10 > tools/quality | grep total
total asprintf usage : 2
total mr_asprintf usage : 1404      /// Much better
total sprintf usage : 55            /// the last one are the more tricky
total snprintf usage : 6
total strcat usage : 9              /// still some more work
total strncat usage : 0
total strcpy usage : 190            /// Much better as well
total strncpy usage : 10
total fgets usage : 77              /// next on the list
total malloc usage : 92
total mr_malloc usage : 25
total malloc_string usage : 109     /// still some more work
total getline usage : 8
total mr_getline usage : 9
total MAX_STR_LEN usage : 185       /// same here
total getcwd usage : 13
total goto usage : 9
total free usage : 33
total mr_free usage : 1429
total paranoid_free usage : 233
total paranoid_system usage : 65
total mr_system usage : 0
total assert usage : 281

That work as originaly done in trunk. But I realized I made mistakes,
and wanted to integrated during the dev of the stable branch rather, so
started over during 2.2.9, and really pushed in Aug. in order to have a
fully dynamically allocated memory management as soon as possible.

So indeed, excellent vacation from all points of view ;-)
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Laurent GEHLEN said on Fri, Aug 21, 2009 at 04:15:18PM +0200:

> > >    - Pass by a cat:
> > > >    $LVMCMD vgdisplay $current_VG | cat > $VG_info_file
> >
> > Ok. I added that in current 2.2.9 SVN as a workaround. Thanks !
> 
> Thanks for it.

Well, was just waiting for your feedback. Now commited in rev [2342]
http://trac.mondorescue.org/changeset/2342
Will be in 2.2.9 final.

> I had already think about it... in theory, changing the context (chcon)
> shoud work. But in actually, it fails because it's not allowed:

Ah, OK, I missed that part of your original mail, sorry.

> > [root@ls003962 /]# chcon -Rt lvm_t /tmp/testMindiSELContext/
> > chcon: failed to change context of /tmp/testMindiSELContext/ to
> > root:object_r:lvm_t: Permission denied
> 
> It's what I was writing in my note just here up, the LVM's Policy give a
> permition denied. I pasted the alert of SELinux here under.

Hummm. Should we open a Bug Report for that for Fedora/RHEL ?

> >     altogether. Disabling SELinux protection is not recommended. Please file a
> >     http://bugzilla.redhat.com/bugzilla/enter_bug.cgi against this package.

Maybe I should no ?

Bruno.
-- 
Linux Profession Lead EMEA  / Open Source Ambassador \   EMEA CME Sol. Center
http://www.mondorescue.org / HP/Intel Solution Center \  http://hpintelco.net
Des infos sur Linux?  http://www.HyPer-Linux.org      http://www.hp.com/linux
La musique ancienne?  http://www.musique-ancienne.org http://www.medieval.org

Bruno Cornec wrote:
> Hugo Vanwoerkom said on Sat, Aug 08, 2009 at 05:38:08PM -0500:
>> GetInitrdFilesystemToUse() looks for the phrase '<6>checking if image is 
>> initramfs...|<6>Unpacking initramfs...' to see if initrd uses an 
>> initramfs fs. *That* phrase is absent.
>>
>> But when I change the phrase to '<6>Trying to unpack rootfs image as 
>> initramfs', which *is* present, mindi successfully terminates and 
>> mondoarchive completes.
> 
> Thanks a lot Hugo for your report. I updated mindi accordingly. Will be in
> 2.0.7.
> 

Bruno, hope you had a great vacation!

Hugo

Bonjour Bruno,

2009/8/18 Bruno Cornec <Bru...@hp...>

> Bonjour Laurent,
>
> >    - Set the SELinux status as disabled or Permissive
>
> Yes. But I'm not happy with that. SElinux *is* useful.


I'm totally agreed. I would really appreciated to not change de status of
SELinux before and after running mindi.


> >    - Pass by a cat:
> > >    $LVMCMD vgdisplay $current_VG | cat > $VG_info_file
>
> Ok. I added that in current 2.2.9 SVN as a workaround. Thanks !


Thanks for it.

> Note that change the context of $MINDI_TMP directory is not possible
> > accounding the SELinux policy for LVM. Because a policy to allow output
> of a
> > confined domains to all possible file_types is not good security or
> policy.
>
> Hummm. Maybe we could just change the attributes of the MINDI_TMP dir to
> allow lvm commands to write in it no ?
>
> > > Additional Information
> > > Source Context                root:system_r:lvm_t:SystemLow-SystemHigh
>
> So if we tag our destination dir with lvm_t, it should wiork no ?


I had already think about it... in theory, changing the context (chcon)
shoud work. But in actually, it fails because it's not allowed:

[root@ls003962 /]# ls -Zald /tmp/testMindiSELContext/
> drwxr-xr-x 2 root:object_r:tmp_t              root root 4096 Aug 21 15:57
> /tmp/testMindiSELContext/
> [root@ls003962 /]# chcon -Rt lvm_t /tmp/testMindiSELContext/
> chcon: failed to change context of /tmp/testMindiSELContext/ to
> root:object_r:lvm_t: Permission denied
>

It's what I was writing in my note just here up, the LVM's Policy give a
permition denied. I pasted the alert of SELinux here under.

And report if this fixes it for good ?

So, unfortunatlly not.


> Thanks for your contribution,
> Bruno.


You're welcome. Glad to do it. ;-)
Laurent


SELinux alert genereted on the chcon:

> Aug 21 16:04:36 ls003962 setroubleshoot:      SELinux is preventing
> /usr/bin/chcon (unconfined_t) "relabelto" to testMindiSELContext
> (lvm_t).      For complete SELinux messages. run sealert -l
> 8c3aea82-1624-4298-9221-0da43a7608d1
>
> [root@ls003962 ~]# sealert -l 8c3aea82-1624-4298-9221-0da43a7608d1
> Summary
>     SELinux is preventing /usr/bin/chcon (unconfined_t) "relabelto" to
>     testMindiSELContext (lvm_t).
>
> Detailed Description
>     SELinux denied access requested by /usr/bin/chcon. It is not expected
> that
>     this access is required by /usr/bin/chcon and this access may signal an
>     intrusion attempt. It is also possible that the specific version or
>     configuration of the application is causing it to require additional
> access.
>
> Allowing Access
>     Sometimes labeling problems can cause SELinux denials.  You could try
> to
>     restore the default system file context for testMindiSELContext,
> restorecon
>     -v testMindiSELContext If this does not work, there is currently no
>     automatic way to allow this access. Instead,  you can generate a local
>     policy module to allow this access - see http://fedora.redhat.com/docs
>     /selinux-faq-fc5/#id2961385 Or you can disable SELinux protection
>     altogether. Disabling SELinux protection is not recommended. Please
> file a
>     http://bugzilla.redhat.com/bugzilla/enter_bug.cgi against this
> package.
>
> Additional Information
>
> Source Context
> root:system_r:unconfined_t:SystemLow-SystemHigh
> Target Context                root:object_r:lvm_t
> Target Objects                testMindiSELContext [ dir ]
> Affected RPM Packages         coreutils-5.97-12.1.el5 [application]
> Policy RPM                    selinux-policy-2.4.6-104.el5
> Selinux Enabled               True
> Policy Type                   targeted
> MLS Enabled                   True
> Enforcing Mode                Enforcing
> Plugin Name                   plugins.catchall_file
> Host Name                     ls003962
> Platform                      Linux ls003962 2.6.18-53.el5 #1 SMP Wed Oct
> 10
>                               16:34:19 EDT 2007 x86_64 x86_64
> Alert Count                   2
> Line Numbers
>
> Raw Audit Messages
>
> avc: denied { relabelto } for comm="chcon" dev=dm-3 egid=0 euid=0
> exe="/usr/bin/chcon" exit=-13 fsgid=0 fsuid=0 gid=0 items=0
> name="testMindiSELContext" pid=2964
> scontext=root:system_r:unconfined_t:s0-s0:c0.c1023 sgid=0
> subj=root:system_r:unconfined_t:s0-s0:c0.c1023 suid=0 tclass=dir
> tcontext=root:object_r:lvm_t:s0 tty=pts0 uid=0
>