Planet CentOS

Dag Wieërs: Improved RHEL, CentOS and Scientific Linux hardware support

Thu, 02 Jul 2009 20:51:28 +0000

The past few months a bunch of very cool CentOS community members (and I am not including myself here) spend their time creating kernel modules for the stock RHEL5 2.6.18 kernel to extend hardware support. The result is now known as: ELRepo

What does it mean ?

It means that if you have problems to get specific hardware working on RHEL5, CentOS-5 or Scientific Linux 5, you can visit http://elrepo.org/ and download kmod packages for your hardware.

It effectively means that with recent hardware (laptops, desktops), you have a high chance that you can run an Enterprise Linux distribution to get recent sound hardware, webcams, dvb-t, file systems drivers and others, to work with no fuss.

It also means that if you require vendor drivers for your server hardware, you can find the latest from ELRepo.

Why should I use ELRepo ?

If you have no problems, you shouldn't. But if you can't make your atheros/ath5k work, ELrepo has a solution. If you need ndiswrapper, just get kmod-ndiswrapper. Same for fuse, ntfs, kqemu, jfs, and many others.

Especially for sound drivers, webcam support, dvb-t, TV mpeg encoders and many more, ELRepo offers the latest drivers backported to RHEL5, CentOS-5 and Scientific Linux 5. Great for using a RHEL5-based solution as a media system. Or when you simply want to video-skype with your girlfriend.

Is it safe to use ?

Since ELRepo only focuses on hardware support, it should work with other repositories quite well. And even when you enable ELRepo, it will not replace existing packages. You pick the kmod packages you need, and you won't get more than that. Remove it, and it's gone.

What about RPMforge DKMS packages ?

In the past I found DKMS the best solution for providing additional kernel modules, but DKMS requires a lot of dependencies, compilation and slows down the boot-process or update-process. Since RHEL5 has a stable kernel ABI there is no real need for all this, and the ELRepo kmod's simply work.

So why still provide those DKMS packages, when ELRepo is so much easier ? Good question :-) That is why I converted all my RPMforge DKMS packages to kmod packages and added them to ELRepo.

In the end, the availability of one big repository providing every kernel module under the sun, with the promise of adding whatever kernel module that makes sense, is a huge step to making RHEL5 or CentOS-5 an even more viable desktop alternative.

I am using CentOS-5 now for 2 years on my laptop (first a Thinkpad T43, now a Thinkpad X200s) and I don't need anything else. Everything works.

PS Yes, I lied. The fingerprint reader has no Open Source drivers (yet), but despite it being very cool, I don't really need it.

Dag Wieërs: RHEL 5.4 beta released

Thu, 02 Jul 2009 00:58:49 +0000

At the very same day as Firefox 3.5 hit the mirrors, Red Hat released Red Hat Enterprise Linux 5.4 Beta with a lot of interesting new functionality:

KVM hypervisor now supported (backported to stable 2.6.18 kernel)
Fencing agents for IBM LPAR HMC and Cisco MDS SAN switches
Updated ALSA drivers (backported to stable 2.6.18 kernel)
Improved laptop docking support
Updated graphics drivers
FUSE support (already available from RPMforge and elrepo.org)
Lots of kernel and driver improvements (too many to list them)

Apart from this Red Hat also shared Technology Previews within each release. Functionality that is not yet supported, but available for testing purposes:

FCoE (Fibre Channel over Ethernet)
More kernel tracepoints for systemtap
ext4 filesystem support
gcc 4.4 and new glibc malloc
CIFS DFS support
samba3x and ctdb support for use with GFS
SPICE X11 drivers (future feature-rich display protocol)
Clustering with KVM hypervisor
Power LPAR support

If you have Red Hat Network entitlements you should have access to the RHEL 5.4 Beta and now is the time to report any issues you may find !

If you are looking for the RHEL 5.4 Beta ISO images, go to Red Hat Network, log in, click on Download Software, expand the RHEL5 channel for your architecture and go to the Beta channel. There you can find the RHEL 5.3 Beta ISO images.

Enjoy !

PS The newer 2.6.18-155.el5 kernel can be download without needing RHN access from: http://people.redhat.com/dzickus/el5/ So test it and report any issues to Red Hat directly !

Karanbir Singh: Connection from UDP log messages

Wed, 01 Jul 2009 10:20:29 +0000

Plenty of people seem to have an issue with snmpd logging connection information for each poll on CentOS 5, like this:

Jul  1 09:50:04 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:59768
Jul  1 09:50:05 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:56329
Jul  1 09:50:05 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:42126
Jul  1 09:50:05 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:47950
Jul  1 09:50:05 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:36634
Jul  1 09:50:05 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:52677
Jul  1 09:50:05 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:44864
Jul  1 09:50:05 doghouse snmpd[4159]: Connection from UDP: [10.0.1.10]:54498

This isnt necessarily a bad thing, however if you want to turn that off, on a fully updated CentOS-5 machine you can change /etc/sysconfig/snmpd.options to something like this :

# snmpd command line options
OPTIONS="-LS4d -Lf /dev/null -p /var/run/snmpd.pid"

Thats about all there is to it. Also, the reason why one finds so many different ways or achieving the same thing on the internet, isn't because everyone is wrong. Its mostly since the -L options syntax and use has been changing over the last few years. And while I am sure there is a good reason for this change but the developers really should consider keeping some backward compatibility in place.

- KB

Original post.

Dag Wieërs: Leaving CentOS team, not CentOS community

Sat, 27 Jun 2009 23:35:56 +0000

With disappointment and regret I decided to resign from the CentOS team after having spent the prior weekend thinking it through. It was not the first time I was in this situation, but this time the number of reasons weighed up against the belief that I can make a difference from within the team.

One month ago, being in the same situation and discussing this with some team members/friends, I regained hope that we could fix this and incidentally, that's when I started the CentOS Pulse newsletter to improve the project's communication.

However a recent incident showed yet again a lack of trust and appreciation and made me question my involvement. There are also a few reasons that I consider internal kitchen and therefore will not be disclosed, but the project lacks leadership, shows poor communication, little transparency and fails to engage more people from the community (eg. great need for improvements, plenty offerings, but slow or little acceptance).

I never had any responsibility in producing the base CentOS distribution and therefore my decision will not have any effect on the final product that people are using. That what makes CentOS a good Linux distribution with, quite possibly, the largest installed Linux base.

And while I may no longer be part of the CentOS team, I still am an avid CentOS user and advocate. And hope that the team (of which I consider many good friends) can turn this around for the benefit of its community, including myself :-)

Fabian Arrotin: Controlling your OOimpress presentations over bluetooth

Mon, 22 Jun 2009 14:05:37 +0000

One other thing I learned from the Florian’s talk last week-end is anyRemote . It can be used to control your Linux laptop (or the application started on your Linux laptop/desktop) , like for example OpenOffice Impress from your mobile phone (over IR/bluetooth/WiFi) . Of course that’s not the only stuff that you can use for that : Dag recently posted his WiiPresent package he wrote during the last Fosdem (co-authored with Didi) but in my case it’s difficult to justify to my kids that ‘Daddy has to steal one of your wiimote’s because he wants to use it during an OSS presentation’ . Advantage of anyRemote is that it’s compatible with my Nokia mobile phone so I was interested in testing/using it. It was not available on RPMforge .. until now ! : i’ve made a commit to the rpmforge svn yesterday (so expect the packages to appear in some days, when Dag’s buildsystem will process them)

People in the meantime who don’t want to wait can ‘ping’ me for the locally built RPMs for CentOS 5

Fabian Arrotin: Interested in Heartbeat/Pacemaker newer rpms ?

Sat, 20 Jun 2009 10:10:23 +0000

While I am/was attending a Zarafa summercamp for professional reasons, I discussed with Florian Haas (from Linbit/DRBD) about newer Heartbeat/Pacemaker packages landing or not in the CentOS Extras repositories (we didn’t talk about DRBD itself which is already provided in the Extras repository while newer DRBD packages are actually in the [testing] one). That’s true that I’ve myself not used/deployed heartbeat based cluster the last months (RHCS instead …) so I didn’t follow what happened on the Linux-HA/Pacemaker level. (I was just aware of the fact that Pacemaker was a replacement for the included Cluster Resource Manager within heartbeat 2.x). The actual heartbeat packages in the CentOS Extras repository were being packaged/built by Johnny but I’ll probably have a look with Ralph about what we can do. Florian told me that he was using the RPMs built by the Novell/OpenSUSE buildservice. While I was following some interesting talks, I had a quick look to see if their SRPMS could be used ‘as-is’ and submitted to Mock. Unfortunately no. I (we ?) ‘ll have to do some cleanups/adjustements within the SPEC file to fit the Mock buildsystem. (OpenSUSE uses someting different of course)

But even if the packages built succesfully , some testing will of course need to be done to see if upgrading from the actual heartbeat 2.1.3 package to 2.99 can be done in a ’smoothly’ way .. More informations to come in (i hope) a near future now that Florian gave me extra-pressure on my shoulders

Fabian Arrotin: VMware : a “time machine” ?

Thu, 18 Jun 2009 03:25:29 +0000

I had (for professional reasons) to look at the Compatibility List on the VMware website for the (recently released) vSphere 4 product. And the the surprise : they already list CentOS 5.4 , even if not even in ‘alpha’ stage .. interesting

Is VMware a kind of ‘time machine’ ?

Dag Wieërs: What inside Firefox is using my CPU ?

Wed, 17 Jun 2009 23:54:52 +0000

I love Firefox a lot, more than any other browser actually. With my small set of addons, me and Firefox, we conquer the world every day !

That's why I hate it when my buddy Firefox is chewing away too many CPU cycles while not doing anything. It is really frustrating because it heatens my lap(top), flattens my battery and slowens my work. (yeah yeah, I know that is not proper English)

I tried all the usual suspects, Google mail and calendar, the obvious heavy website. I closed many tabs, but I never seem to find the culprit. Whatever I end up closing, 100% of one CPU remains going down the drain. Restarting Firefox is no solution either.

Anyhow, I discovered that when this happens, it always is Javascript related. Because the trick is to disable Javascript. (And then simply enable it again because I am an addict !)

So why is there no simple addon that can tell me which of my 100+ tabs is only taking and never giving ? And why, if it does exist, did nobody notify me of this addon ? :-)

Let's hope Firefox 3.5 fixes the problem and not simply uses 100% CPU more efficiently ;-)

Hint: My favorite addons are: Flashblock, Tab Mix Plus and Uppity. I don't need much more...

Dag Wieërs: Community newsletter, take two

Tue, 16 Jun 2009 21:54:02 +0000

I knew exactly what I signed up for when I got this crazy idea to create a CentOS newsletter, still the second issue was less well planned than I had hoped :-/ But then again I am very happy with the result and step by step I am sure things will go easier.

Publishing and announcing it was now a matter of minutes. Geerd-Dietger was of great help to get it to the finish line (in time) ! And the result of the hard work is now available here:

http://wiki.centos.org/Newsletter/0902

Let's try to do even better next edition. But most of all: happy reading !

Karanbir Singh: Gpg signed spam

Mon, 15 Jun 2009 13:10:38 +0000

I've just had my first ever gpg signed spam email. Yes it is gpg signed, and its signed correctly with a key that seems to not be mentioned anywhere ( or atleast papa google does not know about it ). These guys are still getting better and wiser. Whats next ?

Going to see how this shapes up over the next few days before dropping my gpg-header -> whitelist rule :/

Original post.

Karanbir Singh: Thread summaries in thunderbird 3

Sat, 13 Jun 2009 23:52:05 +0000

One new feature I really like in the upcoming Thunderbird 3 tree : thread summaries in the message view pane. Here is an example:

Click the image above for a full size view.

What makes it even more cool, is that since Thunderbird can already recognise and mark the quoted portion in followup replies, it conveniently ignores this quoted text in the summaries.

Original post.

Karanbir Singh: dstat and disk device names

Fri, 12 Jun 2009 15:04:37 +0000

Running dstat normally gives you something like this :

# dstat --nocolor
----total-cpu-usage---- -dsk/total- -net/total- ---paging-- ---system--
usr sys idl wai hiq siq| read  writ| recv  send|  in   out | int   csw 
  3   0  97   0   0   0| 820k  456k|   0     0 | 800B  866B|1054   255

And you can split the disk metrics up based on devices using something like this :

# dstat -D sda,sdb,total --nocolor
----total-cpu-usage---- --dsk/sda-- --dsk/sdb-- -dsk/total- -net/total- ---paging-- ---system--
usr sys idl wai hiq siq| read  writ: read  writ: read  writ| recv  send|  in   out | int   csw 
  1   1  97   0   0   0| 176k   77k:  30k  162k: 411k  478k|   0     0 |3608B 4005B|1447  1458 
 33   2  63   0   1   2|   0     0 :   0   216k:   0   432k|2470k 1611k|   0     0 |2915  6967 
 31   2  65   0   1   1|   0     0 :   0     0 :   0     0 |2210k 1338k|   0     0 |2866  6051

But you can also get deeper level device names / nodes to measure - which is required in many cases if you have drivers for storage creating nodes further down the tree than /dev. Eg. I have a setup where there are four mysql instances running, each with its own dedicated storage :

# mount | grep srv
/dev/cciss/c0d1 on /var/lib/mysql/node1 type ext3 (rw)
/dev/cciss/c0d2 on /var/lib/mysql/node2 type ext3 (rw)
/dev/cciss/c0d3 on /var/lib/mysql/node3 type ext3 (rw)
/dev/cciss/c0d4 on /var/lib/mysql/node4 type ext3 (rw)
/dev/cciss/c0d5 on /srv/wal type ext3 (rw)

so in this case, to get dstat reporting working you need to mention just the component level, like this :

# dstat -D cciss/c0d1,cciss/c0d2,cciss/c0d3 --nocolor
----total-cpu-usage---- dsk/cciss/c dsk/cciss/c dsk/cciss/c -net/total- ---paging-- ---system--
usr sys idl wai hiq siq| read  writ: read  writ: read  writ| recv  send|  in   out | int   csw 
  3   0  97   0   0   0| 209k   89k: 201k   86k: 204k   85k|   0     0 | 799B  865B|1055   256 
 90   1   8   1   0   0|3196k   68k: 764k    0 : 816k   24k|1067k 1650k|   0     0 |1999   609 
 95   1   3   1   0   0|2548k    0 :   0  4084k:2448k 5700k| 660k  791k|   0     0 |1611   571 
 96   1   2   1   0   0|2628k    0 : 808k    0 :1620k    0 | 352k  798k|   0     0 |1835  1605

And you get the details, for each block device.

- KB

Original post.

Fabian Arrotin: “cpio: MD5 sum mismatch” error when submitting a F11 SRPM to Mock

Mon, 08 Jun 2009 20:22:47 +0000

Today I had to migrate a customer CVS repository to Subversion. I looked after cvs2svn but I only found it (at least a ‘working’ version) in Rawhide. “No problem ! , I’ll use my mock wrapper script on my build system” … Except that instead of building a nice ready-to-go rpm, I ended with that error message :

warning: /builddir/build/originals/cvs2svn-2.2.0-2.fc11.src.rpm: Header V3 RSA/SHA256 signature: NOKEY, key ID d22e77f2
cvs2svn ##################################################
error: unpacking of archive failed on file /builddir/build/SOURCES/cvs2svn-2.2.0.tar.gz;4a2d65a4: cpio: MD5 sum mismatch
Error installing srpm: cvs2svn-2.2.0-2.fc11.src.rpm

Hmm, that famous problem Russ reported some time ago . But instead of setting up a F11/Rawhide domU somewhere just to extract the sources/spec from the rawhide srpm , I just decided to modify my wrapper script around mock on my CentOS 5 builder. Instead of just downloading the SRPM and directly submit it to mock, I first install it with the –nomd5 rpm parameter (using rpm2cpio is also an alternative), and then recreate directly a SRPM with `rpmbuild -bs –nodeps` (with of course the correct –define ‘ ‘ values for my build system) and then submit the resulting srpm to mock. I’ll check later if it’s possible to find a rpmmacro that can be used directly in the mock config file to bypass the srpm explode/recreate step. More informations about that issue in the Red Hat bugzilla and also on the Fedora wiki …

Russ Herrold: Phat pipes

herrold@centos.org (herrold -- orc_orc) — Mon, 08 Jun 2009 14:12:00 +0000

Check the top row, right entry ... peaking at 44 megaBytes per second, and a lesser rate sustained over 8 hours; all relevant filtering bridges, and servers in the transfer at our end are running ... CentOS

We've spent the last couple of months in the buildout of our (new) presence in the North data center. We have sites in the central city, on the Dublin fiber ring, and through the north end AT&T switching center, but each has had its faults over time. The downtown 'carrier hotel' was offline for four hours due to a lack of redundancy in its generators during last September's multi day power outage; the Dublin fiber ring peering exchange point had issues as well, but longer; our multi-site strategy saved the day as none of our customers lost inbound data nor went dark in their web presence; uplinks were not affected as we handle them over different routes. In the last couple weeks, AT&T's congestion issues have re-appeared at their plant as well when we were 'babysitting' a large CAD/FEA file transfer ... again multi-gig

The new data center is pricey -- but in addition to the care at the physical layer, it is BGP multi-homed and has really fat pipes. The screenshot up top shows the inbound consumption on the green. Iniitally we had a hard cap on our switch to limit it to 10 MegaBytes/Sec inbound -- but we were doing a large (a multi hundred gigabyte pull), and dropped the cap once it was clear all was working well

We are in the paperwork phase at the moment with ARIN, to clear up some 'lint' on our ASN, but with any luck by the end of the month, we'll have completed the cutover

Fabian Arrotin: An opensource backend to sync my mobile phone

Fri, 05 Jun 2009 07:54:52 +0000

While I used for several months the service offered by ScheduleWorld , I didn’t like the idea that my calendar was stored elsewhere than on one of my machines. The fact that ScheduleWorld decided recently to switch to V2 (and now don’t provide the service for free anymore), it pushed me to find a solution to sync my calendar between my Nokia E51 and my Linux laptop/computers. I really appreciate my Nokia mobile phone, but unfortunately it doesn’t support iCal (and I’ve not found a symbian app that could do that ..) . The only protocols that the Nokia can ‘talk’ is SyncML or ‘ActiveSync’ (through their ‘Mail for Exchange‘ free plugin) . That directly limits the scope for the backend. While I considered Funambol at a time (to use SyncML) , I finally ended with Zarafa (and Z-push) . It’s all open-source (in the community edition though) and emulates an ical (and caldav support is now available in the 6.30 release) and Z-push emulates an ‘ActiveSync-over-the-air’ server so I’m now able to directly sync my calendar/contacts/tasks/mails from my Nokia mobile phone to the server (using a MySQL backend) and either use the Zarafa webaccess (that I don’t use that much though) or Thunderbird with the Lightning extension . (every “iCal aware” program works of course)
Note : Z-push isn’t yet available in the RPM format on the Zarafa website due to a clause in the GPL license (more informations on the RPMfusion bugzilla related page) . Thanks to Robert Scheck a spec file was written but isn’t yet available. Robert is interested in seeing his package landing in EPEL and RPMfusion while I consider myself providing it in RPMforge. In the meantime, if you’re interested in the RPM version, feel free to ‘poke’ me or consult the spec file in the RPMfusion bugzilla .

Dag Wieërs: CentOS Pulse - Community newsletter

Tue, 02 Jun 2009 15:35:09 +0000

A CentOS newsletter was what I wanted to have for a long time. It finally materialized the past week and was published today. You can read it at:

http://wiki.centos.org/Newsletter

We plan to release a new issue every 2 weeks and for the time being I will be the interim editor. The foreword explains why the newsletter is a new starting-point for the CentOS community and hopefully we can use it to communicate decisions, needs, hopes and events better and in a transparent way.

But most importantly the newsletter should voice the opinion of the community. I am a long-time Linux Weekly News follower and I hope we can set the same standards with our newsletter as LWN, not from the very first issue, but hopefully as we grow into making it, step by step.

Fabian Arrotin: RPM the easy way ?

Sat, 30 May 2009 08:26:26 +0000

While Karanbir posted an interesting rpm the other day , that reminded me another commercial app I had to look once. The application was provided as an RPM, but it seems that none of the installed files was declared in the rpmdb .. and here is why :

[arrfab@waldorf vmware]$ echo -e “Files present in the RPM package: \n” ; rpm -qlp VMware-Player-2.5.1-126130.x86_64.rpm ; echo -e “\nand now the RPM script : \n” ; rpm -qp –scripts VMware-Player-2.5.1-126130.x86_64.rpm
Files present in the RPM package:

/var/cache/vmware/VMware-Player-2.5.1-126130.x86_64.bundle

and now the RPM script :

preinstall program: /bin/sh
postinstall scriptlet (using /bin/sh):
# Execute bundle installer on install or upgrade after laying down bundle
# and then delete the bundle afterwards.
# Have to redirect the console to stdin because it’s closed by default.
# Setting VMWARE_SKIP_RPM_UNINSTALL is necessary because we don’t want the
# bundle to run rpm commands, since rpm will deadlock if that happens.
TERM=dumb VMWARE_SKIP_RPM_UNINSTALL=1 /var/cache/vmware/VMware-Player-2.5.1-126130.x86_64.bundle \
–required –console /dev/tty
rm -f /var/cache/vmware/VMware-Player-2.5.1-126130.x86_64.bundle
preuninstall scriptlet (using /bin/sh):
# On uninstall only, remove existing bundle installation.
if [ $1 -eq 0 ]; then
if [ -e /usr/bin/vmware-uninstall ]; then
TERM=dumb /usr/bin/vmware-uninstall --console > /dev/null 2>&1
fi
fi
postuninstall program: /bin/sh

Do we really have to comment on that one ?

Karanbir Singh: Getting yum to show all packages that match

Thu, 28 May 2009 17:40:58 +0000

One issue that lots of people run into is when they want a list of all pacakges in the repositories that match their requirements. By default yum will only show you the 'newest' or the 'highest version'. So looking for kernel-devel does this:

[root@chamkaur ~]# yum list kernel-devel
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
 * fasttrack: mirror.ukhost4u.com
 * extras: mirror.ukhost4u.com
 * updates: mirror.ukhost4u.com
 * base: mirror.ukhost4u.com
 * addons: mirror.ukhost4u.com
Installed Packages
kernel-devel.x86_64                2.6.18-8.1.10.el5                   installed
kernel-devel.x86_64                2.6.18-53.1.4.el5                   installed
kernel-devel.x86_64                2.6.18-92.1.10.el5                  installed
Available Packages
kernel-devel.x86_64                2.6.18-128.1.10.el5                 updates

Thats all good, except we have :

[root@chamkaur ~]# uname -r
2.6.18-128.1.6.el5

So what we are looking for is kernel-devel that matches our running kernel. Now, one way to do this is to run 'yum list kernel-devel-2.6.18-128.1.6.el5'; and if its there, we can install it. However, yum also has option called '--showduplicates', which will list all matches, not only the latest. Trying yum with that option gives us:

[root@chamkaur ~]# yum --showduplicates list kernel-devel
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
 * fasttrack: mirror.ukhost4u.com
 * extras: mirror.ukhost4u.com
 * updates: mirror.ukhost4u.com
 * base: mirror.ukhost4u.com
 * addons: mirror.ukhost4u.com
Installed Packages
kernel-devel.x86_64            2.6.18-8.1.10.el5                      installed 
kernel-devel.x86_64            2.6.18-53.1.4.el5                      installed 
kernel-devel.x86_64            2.6.18-92.1.10.el5                     installed 
Available Packages
kernel-devel.x86_64            2.6.18-53.1.19.el5.bz321111            c5-testing
kernel-devel.x86_64            2.6.18-92.1.1.el5.bz444759             c5-testing
kernel-devel.x86_64            2.6.18-92.1.6.el5.bz_pre53             c5-testing
kernel-devel.x86_64            2.6.18-128.el5                         base      
kernel-devel.x86_64            2.6.18-128.1.1.el5                     updates 
kernel-devel.x86_64            2.6.18-128.1.6.el5                     updates 
kernel-devel.x86_64            2.6.18-128.1.10.el5                    updates

Which gives us all the various versions of kenel-devel visible to yum, in enabled repos for the machine. And we can then go ahead run 'yum install kernel-devel-`uname -r`; and know its going to pull in the right version. The --showduplicates is also a good way to check the history of a package during its release cycle. And this option should become a lot more useful when you have the vault repos enabled.

Original post.

Karanbir Singh: Checking a machines yum exclude policy

Thu, 28 May 2009 17:27:51 +0000

Want a quick and easy way to check what the Excludes policy on a machine is ? Run this :

wget -q http://centos.karan.org/check_excludes.py -O - | python

It should give you some nice clean output, listing each repository enabled on the machine, along with the exclude setup for that repo. Something like this :

main: []
centos-base - ['perl', 'rsync']
centos-updates - ['perl', 'rsync']

Remember that the "main:" section in yum is the global system/repowide configs - anything listed there would be excluded from all repositories.

I needed it, but its Seth who wrote this snippet of code.

Original post.

Karanbir Singh: Removing lots of data from a DB table

Tue, 26 May 2009 14:30:44 +0000

If you need to remove large amounts of data from a table ( specially when you need to remove *all* the data in a table), remember its faster to drop the table, and recreate it rather than doing a 'delete from tablename;'.

eg: this is against a postgresql db:

drop table rawdata;

Response time: 0m 2s 433ms  Total time: 0m 2s 433ms

CREATE TABLE "rawdata" ("ip" char(33), "sdate" varchar(20) NOT NULL,
 "key" varchar(100) NOT NULL,"country" varchar(2), "region" varchar(2),
  "city" varchar(40), "machine" int);

Response time: 0m 0s 337ms  Total time: 0m 0s 337ms

Whereas:

delete from rawdata;
200000000 row(s) affected

Response time: 0m 15s 353ms  Total time: 0m 15s 353ms

In some cases, where you want to drop most ( eg > 80% ) of the data in the table, its faster to select the data to be retained into a temp table, drop the original table and rebuild the temp table to the original name.

- KB

Original post.

Dag Wieërs: Gnome thumbnails in CentOS

Sat, 23 May 2009 08:12:58 +0000

Since some time I was wondering why my Gnome did not generated thumbnails for various video formats (avi, mkv, ...) as thumbnails in my file browser (nautilus) really help me navigate.

Today I investigated some more, and even though I have the whole gstreamer-plugins set installed what I apparently needed was gstreamer-ffmpeg !

When RPMforge is enabled, simply do:

apt-get install gstreamer-ffmpeg
or
yum install gstreamer-ffmpeg

Once installed, I still had to remove the "failed" thumbnails (a placeholder so nautilus does not try to rebuild thumbnails for files it failed before). You can simply do this by removing the hashes:

rm -vf ~/.thumbnails/fail/gnome-thumbnail-factory/*

Then just open the folders to regenerate those thumbnails.

Karanbir Singh: say what Arch is this package

Wed, 20 May 2009 15:10:02 +0000

# rpm --qf "%{name}.%{arch}\n" -qp hpacucli-8.0-14.noarch.rpm
hpacucli.i386

# rpm -qpi hpacucli-8.0-14.noarch.rpm 
Name        : hpacucli                     Relocations: (not relocatable)
Version     : 8.0                               Vendor: HP Company
Release     : 14                            Build Date: Tue 05 Feb 2008 07:37:37 PM EST
Install Date: (not installed)               Build Host: localhost.localdomain
Group       : Applications/System           Source RPM: hpacucli-8.0-14.src.rpm
Size        : 12596699                         License: 2005 Hewlett-Packard Development Company, L.P.
Signature   : (none)
Summary     : HP Command Line Array Configuration Utility
Description :
The HP Command Line Array Configuration Utility is the disk
array configuration program for Array Controllers.

Fail!

Original post.

Karanbir Singh: push rather than pull based rsyncs for mirrors

Thu, 14 May 2009 09:57:27 +0000

Has anyone given though to doing mirror updates via a 'push' rather than a 'pull' mechanism ?

Basically, when new packages are available the machines that get updates would do a rsync to the machines down the order, pushing the updates out. It would allow us to get more updates, out faster and also not need to waste cpu and i/o on repeated rsyncs that dont need to be run. It would also make it a lot more viable to run rsyncs with a '-c' option always set.

Yes, we could/would/should still leave in place some mechanism for people to setup new mirrors and also to do pull based rsync's as and when they need.

Just something to think about at the moment, and comments would be very welcome.

- KB

Original post.

Russ Herrold: Rainy Days & Mondays

herrold@centos.org (herrold -- orc_orc) — Mon, 11 May 2009 16:03:00 +0000

Karen Carpenter made the song famous for its authors, but clearly none of them were sysadmins

The rule, long known, for sysadmins is:

Never make a major change on a Friday, nor before leaving for vacation

I've been wrestling with the fallout from a violation of the sysadmin's rule by an upstream provider -- the vendor pushed in some change on Friday in the preparation of CDR -- Call Detail Records. For four days running, my sub-processes which manage the account have been failing for want of data. Those processes retrieve and apply CDR data, to emit accounting detail for customers, and have not been working

I've filed five or six sub-issue tickets which that primary change exposed, in trying to get the matter resolved: The current Firefox cannot open tickets under the current Windows XP, current SP [no problems with CentOS and

FireFox or konqueror]; my 'closed' tickets were not visible; tickets were being closed by upstream before I confirmed a fix worked, so I ended up essentially re-opening the same ticket three times as each day's CDR pull failed; I was not receiving email updates of tickets; and so on. I am quite sure they consider me a 'stickler for details' and something of a pedantic pest at the moment, but dammit, I'm paying their bills. The PHB supervisor may want tickets closed quickly; but I want my issues fixed first

... as no one likes to be called into work on the weekend to revert a change, the sysadmin's rule must be faithfully applied

Russ Herrold: Getting to a x86_64 build environment

herrold@centos.org (herrold -- orc_orc) — Mon, 11 May 2009 15:05:00 +0000

In the #centos IRC channel on freenode, today, a new user was trying to clean out the 'multi-lib' artifacts in his build environment, so that it was only generating 'x86_64' results

Tom mentioned:

10:26 Zathrus> realistically, just removing glibc.i?86 should nuke everything else...

and so I fired up a victim xen instance to test that hypothesis

sudo su - cd /etc/xen ls cp centos-5-x86_64-test centos-5-x86_64-victim joe centos-5-x86_64-victim # the edit is to rename the instance name, and the image to be used cd /var/lib/xen cp centos-5-x86_64-test.img centos-5-x86_64-victim.img xm create centos-5-x86_64-victim virt-viewer centos-5-x86_64-victim

Then inside the instance as root, I ran:

rpm -qa --qf '%{name} \t %{arch} \n' | sort > pre-remove.txt yum remove glibc.i?86 rpm -qa --qf '%{name} \t %{arch} \n' | sort > post-remove.txt grep -v x86 post-remove.txt | grep -v noarch

getting the result:

gpg-pubkey (none) libaio i386 libgcc i386 python-devel i386

That's a pretty good result for a first pass, and a quick hack. I think I'll go down for some coffee, and think about it a bit more

Dag Wieërs: Red Hat backported I/O accounting to RHEL5

Thu, 07 May 2009 18:24:43 +0000

One of the things I do on a weekly basis is follow the kernel development that Red Hat undertakes for their future RHEL5 kernels. This is very interesting because you can check the changelog for fixes, new hardware support, backported features (eg. kvm) and newly supported stuff (eg. fuse, xfs) that is coming in RHEL 5.4.

We discovered xfs was coming to RHEL, new ath5k fixes prove helpful on a friend's laptop, and I was waiting for I/O accounting, kvm and fuse to hit these releases too.

Yesterday I got an early birthday present as Red Hat backported I/O accounting in their 2.6.18-144.el5 kernel and it works really well.

Dstat can show I/O accounting information and the topio and topbio dstat modules show you the biggest I/O consumer both virtually as well as physically.

Virtually in the sense that it includes the I/O that is going to and coming from the pagecache. Physically is what comes and goes from the real disk.

In the past during presentations I always had to demonstrate this from a Fedora image in Virtualbox, but being able to show that on a running laptop is much more convenient because it represents the real workload of a common CentOS-5 desktop system. This is what you would see with dstat on a 2.6.18-144.el5 kernel:

[root@moria ~]# dstat -M topio -d -M topbio
----most-expensive---- -dsk/total- ----most-expensive----
     i/o process      | read  writ|  block i/o process   
rhythmbox    26k:  16k|   0     0 |                      
rhythmbox    30k:  16k|   0    20k|kjournald     0 :4096B
rhythmbox    25k:  16k|   0     0 |                      
rhythmbox    26k:  16k|   0     0 |                      
rhythmbox    30k:  16k| 128k    0 |rhythmbox   128k:   0 
rhythmbox    26k:  16k|   0  8192B|                      
rhythmbox    26k:  16k|   0     0 |                      
rhythmbox    30k:  16k|   0   104k|kjournald     0 :  28k
firefox     225B:  73k|   0   104k|firefox       0 :  80k
rhythmbox    30k:  16k| 128k 4096B|rhythmbox   128k:   0 
gconfd-2      0 :  92k|   0    92k|gconfd-2      0 :  96k
rhythmbox    26k:  16k|   0     0 |                      
rhythmbox    30k:  16k|   0   124k|kjournald     0 :  48k
rhythmbox    26k:  16k|   0     0 |                      
rhythmbox    26k:  16k| 128k    0 |rhythmbox   128k:   0 
rhythmbox    22k:  16k|   0     0 |

Showing the biggest normal I/O consumer (application level, virtual), the total disk statistics, and the biggest real I/O consumer (disk level, physical). rhythmbox is playing music and reading 26k/sec, but in reality it loads about 128k every 5 seconds. kjournald is invoked after something is being written. Firefox is writing to its cache.

And what gconfd-2 is repeatedly doing, I would really like to know because it is my biggest mystery (writing!) consumer :-/

PS: I also need to do something to my Drupal theme, since it is not properly showing consecutive spaces in <code> blocks...

Russ Herrold: May Day celebration

herrold@centos.org (herrold -- orc_orc) — Fri, 01 May 2009 13:30:00 +0000

I see a tweet from KB about the monthly mailman mailing list reminder emails; I took steps long ago to use procmail to watch for these, and re-mark their subject line. I then sort my mailspool by subject in alpine, and delete this noise all in one pass.

# mailing list memberships reminder
#
:0 fw
* ^Subject: \/.*mailing list memberships reminder
*!^X-Reminder
  | formail -i "Subject: mlmr] $MATCH"                \
    -A "X-Reminder:$MATCH"    \
    -A "X-Munge: moved mailing list memberships reminder"

All power to the people

Dag Wieërs: Are there too few people who understand desktop Linux ?

Wed, 29 Apr 2009 00:19:53 +0000

I just read the ComputerWorld article "Are there too many desktop Linuxes?" and once again I think the author is missing something very obvious and important.

If we are talking about Desktop Linux, we are talking about the consumer and enterprise market. If we are talking about the consumer and enterprise market, we are talking mostly about non-technical people and not me or you (since you are reading my blog, I do not consider you the average computer user).

I usually describe the consumer and enterprise market to be 99% of the population, which I often visualize as: my mother !

For those that are already confused, my mother is non-technical and not a power user. She uses applications and often struggles with tasks I consider basic. But this mostly means I too have a skewed perception of computer users and basic computer skills.

(If your mother is too distracting as an example, consider you are a support engineer of a big enterprise and you are in charge of the enterprise desktop systems of the company's management including the CEO.)

A desktop Linux system is a means to a goal, and not the goal itself for 99% of the people.

Now the following list of facts exists and may be inherited in the consumer and enterprise markets:

Technical people have little time or patience to upgrade, maintain or provide support for the 99% of other people. I can do it for a selected few only (in my spare time).
My mother (or your CEO) is only interested that her applications work reliably, she does not need the latest bleeding-edge features, nor does she want new versions of applications or a new OS every 6 months that may look or behave differently to what she is used to.
I am not interested in updating/upgrading my mom's computer whenever I pay a visit because the OS demands that from me. I'd rather do other things with my time.
The longer I can have my mom's computer running (think years) in a secure manner, the fewer worries I have.
Every intervention that can be avoided, is time I can use for something else. Or is money saved for the company.

This means that high-maintenance, bleeding-edge Linux distributions, while very useful to the selected few, are not a nice fit for 99% of the market because they change too often, require work to update them and change often spurs breakage, instability or behavioral change.

"If it ain't broken, don't fix it." applies in the consumer and enterprise market very well (unless maybe when you make money by fixing).

So, we end up dismissing all distributions that:

have less than 5 years of support (security updates)
have a new release (and forced upgrade path) that is less than 3 years
update applications during the course of the lifetime with little consideration to support/stability

Which means most Linux distributions are not suitable as a Desktop Linux for the majority of the population. So not Fedora, not Ubuntu, not OpenSUSE, not any of the other distributions except:

RHEL/CentOS
Ubuntu LTS
SLES/SLED

the Enterprise distributions. And I have nothing against the Fedora/Ubuntu/OpenSUSE/whatnot distributions of today, they are very needed for driving progress and for fostering the Open Source community. But please, not at the expense of the non-technical end-user.

I know that many of you have compelling reasons to like and use bleeding-edge distributions, and that is fine. But please also consider that your requirements are not necessarily the same requirements as my mother or your CEO. And the way you manage your own computer does not scale very well if you need to support 10, 100 or 1000 end-user Linux desktops.

Karanbir Singh: Email Server farkage

Wed, 22 Apr 2009 16:32:54 +0000

My email server decided to go south a few minutes back - and the backup isnt setup to really run in full production mode, which means that I have limited access to emails till the server is back up.

So if you are expecting emails from me, or have emailed me something - patience, I will get back to you but it might be only after early morning on Thursday 23rd Apr.

- KB

Original post.

Fabian Arrotin: Small thoughts about the upcoming RHEV

Wed, 22 Apr 2009 09:37:59 +0000

While I attended the Red Hat partner summit, we had a demo of the upcoming RHEV (for servers and desktops). It was strange that while Vmware announced a beta version of VirtualCenter running on Linux, on the their side, Red Hat decided to keep the version written in .Net (from people from Qumranet, acquired by Red Hat last year). So you need a Microsoft Windows 2003 machine to manage your Red Hat Virtualization infrastructure .. are times changing ?

Of course we know that Red Hat is an opensource company and that each time they acquired a company they opensourced properly the product (Directory Server, GFS, etc …) so we’re sure that the goal is to provide a Linux version in the future .. But due to the fact that all Virtualization companies are now in a race, Red Hat didn’t want (again) to wait several months (even if RHEV ETA is september). Of course we can trust Red Hat on that one .. but on the other hand , Red Hat addicted people were astonished when we saw a Windows machine with Internet Explorer. Something nobody swore it would happen some years ago …