An intermediate 2.1.5 mindi version

It’s not very often that I separate mindi from mondo in the publication of releases. But this time it was needed as I had a customer who suffered from bugs that were only needing a mindi realease, and I thought it would help many other users ,so here you are !

Mindi 2.1.5 is there, and is principally solving kernel support detection for the type of initrd possible (solves an abort of mindi on RHEL3/4), and also reduces the number of error messages when dealing with links containing more than 2 references to .. Should help with some recent reports.

Also I had a report that the -H option and RESTORE keyword were not completely without interaction, so this is now solved as well.

Finally, this version supports better HP ProLiant Gen8 and future platforms by also using hp-rcu and hp-fm tools.

Now available on ftp://ftp.mondorescue.org for more than 120 distribution tuples ! And for those who ask why I do that: first because I like it, then because I have the tools to do it, and also because I do have users who are using Fedora 7, RHEL 3 or even Red Hat 6.2.

MondoRescue 3.0.3 is finally out

I know. It’s been a long process. And I’m not even fully satisfied with the results, as we do have unresolved bugs in this version. But as it also fixes some critical others that were expecting for a long time (3+ months), I didn’t want to delay more the release of this version.

So here it is. mondo 3.0.3, mindi 2.1.4 and mindi-busybox 1.18.5-3 are now availble from the master ftp server. And as given in the announce, most of the distributions (nearly 100 of them) also have packages ready for use. Once more couldn’t do it without project-builder.org

The detailed change logs are available for mindi, mindi-busybox and mondo.

Among the main fixes brought by this version, a much better support of SLES 10, SLES 11 SP1, SLES11 SP2, especially for LVM and device exclusion, grub, keyboard.

I also received very, very valuable contributions from users, which helps producing a better version of course, and reduce the time of the resolution of the problem in the upstream version. Thanks again for helping ! Including on some very old bugs. And Victor being one of the most prolific.

There are even some new features such as the support of swaplabel or the inclusion of all tools part of minimal.conf now in the first boot part.

As said earlier, some bugs ar not fixed yet. I passed most of the week trying to fix the automatic mode where CTRL-ALT-DEL doesn’t work. I added support in 3.0.3 of an inittab file, but whatever the conf I use fo now, it doesn’t trigger a reboot when using the magic key or sending it from the KVM monitor. So be warned and use with care. I’ll report on the busybox ML as I still have the issue with 1.20.2, and try to get help from this community. Could well trigger the release of a new mindi-busybox (and maybe mindi) soon after this one.

In the serie of un-fixed bugs for lack of time for this time are some that a user reported as more important to fix soon:

• Ticket #628: 3.0.2-1 Fail to Restore from External Hard Disk – This is a very big issue here.
• Ticket #641: Specifying the Same Backup to NFS via CLI=Success via GUI=Fail – Although it is possible to work around this problem it is preventing the creation of exceptional backups by users with little knowledge of Mondo. A fix will be most welcome and will present a more appropriate impression of Mondo.
• Ticket #640: mondo.tmp.xxxxx & mondo.scratch.xxxxx Issues – This is not a critical issue. A fix will help with backup management and housekeeping tasks.

I globally agree these are also important to fix (even if less than Ticket 627). I also think we need to work on Ticket 656 as well. I’d like to get your comments as well in order to prioritize the fixes in the next version, as I can’t work full time on MondoRescue, so need to choose where to put my efforts.

busybox fix while preparing MondoRescue for its 3.0.3 version

While preparing the new version of MondoRescue (test packages on their way for final remarks before having next stable version), I was trying to make a missing version of mindi-busybox for a colleague still using RHEL 3. I quickly realized that my promise to make it (hoping to just type pb -p mondorescue -m rhel-3-i386,rhel-3-x86_64 -r tags/3.0.2 cms2vm) was not so easy to perform

It happens that the version we use (1.18.5) doesn’t compile out of the box for such an old beast as a RHEL 3 with it’s 2.4 kernel. So I had to make some adaptations to the code, in order to have it compile for RHEL3, but still working for newest versions. At least, I have something which seems to be working for RHEL 3, RHEL 6 and Fedora 18, so now I’m building for all my supported distros which now are 120+ and will probably take up to mid day tomorrow.

For those of you interested, had issues with the main Makefile, statfs.h (HAVE_SYS_STATFS_H shouldn’t be set in RHEL 3 case) and a missing BLKGETSIZE64 macro, and the corresponding patch is at: http://trac.mondorescue.org/changeset/3085

With that I’m also trying to fix 2 remaining bugs (http://trac.mondorescue.org/ticket/627 and
http://trac.mondorescue.org/ticket/651) which should have no other impact. The first one is still not fully tested as corrected.

My goal is to publish the next stable version 3.0.3 of MondoRescue this week-end, for my birthday

Finally time to publish pb 0.12.2

As you can see with the few messages I had time to post on this blog, the end of 2012 start of 2013 has been pretty busy, and I’m late in delivering the 2 projects I’m leading. So this week, I decided it was time to make a 0.12.2 version of pb, and make it available. Was asked by my colleagues of FOSSology, specially to add Fedora 18 support, so I also build my VMs to make packages on this distro.

BTW, a bit of ranting for Fedora once more: no perl by default , and no ifconfig nor route command either which is breaking MondoRescue of course, and I guess tons of other software around. These guys don’t care about past, but don’t provide compatibility tools either !! So I’ll have to make new patches, just to do the same as what was done, but with another command again. Not to speak of systemd which I still have issues to deal with That doesn’t make stuff go faster !

Anyway, the version is now out, no official bugs fixed, but a lot of small stuff here and there which were desrving a release. No time to test Fedora 17/18 VE yet, so you’ll have to do that yourself if you want. I also had a look at virsh usage in combination with pb, and it’s again not as easy as it could seem to be. Especially port redirection I’m easily using by launching qemu-kvm manually with the -redir or hostfwd option doesn’t seem to be possible with the user mode network through virsh (neither manager nor CLI). Will have to post on their ML to see how they do that, if they can !

And MondoRescue has been very late. I really need to publish a verion, but I still have some blocking bugs I really would like to get rid of: CTRL+ALT+DEL not working anymore during restore, some LVM issues on RHEL, some grub issues on SLES… Hopefully at the end of next week I’ll have made progresses.

Of course patches are much easy to integrate, but I receive more bug reports than them

And also back to preparing the HP internal TES event, Solutions Linux confs, submitting to LinuxCon, working on an FLOSS ITIL stack, learning more OpenStack, looking at Intel’s TXT…. so many things I’d like to do or learn and will never have time to ! Maybe at least I’ll talk about that another time.

Availability of the beta of MondoRescue 3.0.3 (and mindi 2.1.4)

We have had a long work on this version to fix some issues reported by a customer on SLES 11 SP2 particularly, which will benefit all MondoRescue users. The support of that last version of SLES is now fully working, and it allowed us to fix issues around LVM, /dev/dm-* devices and grub boot loader re-installation. A lot of other bugs have been fixed during this version (some 23), and I really encourage you to test it and report your findings on our mailing-list.

Before the final publication, I’d like to fix the issues encountered by Fedora 17 users, but I’ve not neen able to reproduce it correctly on my side up to now, and I stil l have to work on the latest log files reported to me in order to understand what is creating the issue some users are seeing.

But if no blocking point is found before end of week, and if my tests are successful, I think we’ll have a good 3.0.3/2.1.4 couple that I’d like to push as stable. Fedora fixes could then be done in a later mindi version (as this is what is mainly creating a problem now).

Thanks for your patches, support and now tests !

Mindi 2.1.3 is now available

Ok, so before 1 week in my other life (amateur early music musician), I want to publish this new update of mindi only. It fixes 4 minor bugs that have been reported, mostly with fixes, and that should make MondoResuce more robust again. To be used with mondo 3.0.2.

This version brings:

• Fix #621 by avoiding to handle iso9660 and bind FS type from fstab.
• Fix a bug in mindi where regular expressions were used with grep without the -E option
• Fix #614 SLES kbd issue
• Improves Mageia 2 support which needs .ko.xz modules detection

All packages have been made and are available at ftp://ftp.mondorescue.org
So happy disaster recovery !

Last 2 weeks too hot…

Had lots to do during the last weeks. New MondoRescue Version, new project-builder.org version, Solutions Linux event in Paris, new HP internal projects, and on the private side rehearsal + concert and vocal ensemble conducting, finding a room for my daughter next year in Paris for her studies was a bit too much, with not enough sleep.

Was completely tired last thursday, with 39°+, and slept most of Friday as a consequence. Saturday, I still had not completeley recover, Not a real serious burn out, but made me think I was probably in need of rebalancing a bit stuff to avoid a serious one. And it took me nearly a week to recover, even if I dstill don’t feel exactly as before.

So I’ve just restarted this week to work on some mindi patches, and even if I planned to release a new version of mindi during the week-end it will now be during this !

I’ll be on vacation next week, for my yearly renaissance music week near Paris. So if you want to meet, for once, it will be at one of the concert we will perform the 13th, 14th or 15th of July as explained in the link ! Hopefully that change of activity will let me rest a bit a come back stronger

MondoRescue 3.0.2 is now available

I had to make this release in order to use it today (well yesterday now already !) on a customer site. That version was needed to support correctly software Raid on RHEL 6, with re-creation of metadata and UUIDs, which took a bit of time to make correctly. This version will also be welcome by Debian and Ubuntu users, as it fixes restoration issues, as well as on RHEL 5 (tune2fs command wasn’t working for ext4, and it required to use tune4fs, only available here !). It supports now out of tree kernel modules (such as hpsa from the PSP/SPP) and improve support for latest HP ProLiant Blades.

It fixes lots of small annoyances here and there (like #616), especially thanks to a test script that I have now developed to automate regression tests. This version was tested with it for rhel-6-x86_64, rhel-5-x86_64, debian-6.0-x86_64, ubuntu-11.04-i386, fully for the first time. I plan to extend it so that in the future I can cover more cases. And fixes also came from external contributors, which is very nice to see

So, now time to rest a bit ! After project-builder.org and this version of MondoRecue, I can take a bit of time to think to the next steps, and more over, work on some other projects such as the UUWL and some I’ll announce later.

New Mondorescue 3.0.2 beta version available

This version will fix again some problem met by customers or community users. Among these, as detailed earlier, the crash at restore time that was affecting users of MD software raid volumes, and the fact that we are now supporting UUID for their designation such as on RHEL 6.

Also for RHEL 5 users, there was an error on “no space left on device” at restore time, due to a modification in the way we are including more tools in the initrd, leading to the inclusion of the MAKEDEV program which was creating issue on device creation leading to that error message. It turns out this is also triggered by the busybox shell, so as we need anyway bash in the initrd, we will now use bash by default as the shell to launch all our scripts at restore time.

Mindi log files included in mondoarchive.log will also now be the expected one, not the one from an intermediate run.

2 contributors also provided fixes for issues: one for correct keyboard support by SLES by Victor Gattegno (who also helped around other bug fixes in this version), another one around the support of multiple PVs by bzium.

Even if 19 bugs were closed, I still have 2 issues that I’d like to fix before releasing 3.0.2: one around SLES grub install issue, and one around duplicate hpsa driver. And there are some others that may find their way in it. But that will be difficult, as I have a training in UK from the 2nd of May to the 4th of May, so I’ll publish the new version before in order to have it available at the training time !

So please test the beta version I made available at ftp://ftp.mondorescue.org/test and let me know if you find it stable for your case, and what other remaining issues you’d like me to work on next.

MondoRescue and Software Raid

I had the opportunity to work for a customer whose infrastructure is using massively software RAID via md on a SAN storage. Their install base is made of RHEL6 VMs and they wanted to use MondoRescue for their imaging. What else as some could say

Recent versions of RHEL do use UUIDs everywhere, including to address MD devices in grub configuration, or mdadm ones. That doesn’t make the disaster recovery easier, if you recreate the device from scrtch with a new UUID. So the best approach is to store the information at backup and recreate them with the same UUID they had at restore time. But even if UUIDs on filesystems are supported since quite a long time now, it wasn’t the case for MD devices up to recent SVN revisions. In fact multiple issues were found, trying to make this support work correctly, which were gathered in some existing (and old) MondoRescue trac bugs (#73, #473, #500) or some especially raised at this occasion (#595 and #596).

I have now extended one of my test program to add MD tests as well, and it allowed me to finally solve all the remaining issues linked to this support. Hopefully ! In particular, we now also restore correctly the metadata format of the MD device, in order to be compliant with the boot loaders, as not all of them, or their versions, support all metadata versions. Not clear ? Well try to boot on a md device with the 1.2 version of metadata (created by default with latest mdadm create command) and you’ll rapidly understand

So as you could have guessed, the next step is now to produce a new set of packages in order for you to test As usual they will be available under ftp://ftp.mondorescue.org/test/ where you can pick probably your distribution of choice.

But that’s not all what I’d like to fix for the upcoming 3.0.2. I need to look closely at the bug #600 as we have an issue with the latest MondoRescue version on RHEL 5.x where x is recent as well. After I’ve fixed this one, I think we’ll be good to publish 3.0.2 officially, and start chasing other bugs for the next one