Archive for the 'mirror server' Category

Update To Fedora 12

Published on January 11, 2010 in fedora and mirror server. 0 Comments

Last week I have finally updated our mirror server to Fedora 12. It was still running Fedora 10 which has reached its end of life. The server was running Fedora 10 for a long time and it was always running with a CentOS kernel. The Fedora kernels were, at the beginning, not stable enough (crashing after three or four days) so that I quickly switched to a CentOS kernel. I know that I should have reported bugs, but in the case of the mirror server I am more concerned to keep it up and running than getting debug data from it. It also not easy for me to get physically to the machine so that I had a lot of good excuses to switch to a CentOS kernel.

Now the system is running using the Fedora 12 kernel and after a week it is still up without any problems.

Updating My RPM Fusion Builder

Published on January 8, 2010 in adrian and mirror server. 0 Comments Tags: .

I am running one of the RPM Fusion builders in a VM using CentOS and after I saw that the newly created VMs on my notebook are using virtio for network and disk access I thought that I will try this also for my builder VM. It was pretty easy and straight forward.

First I had to update from CentOS 5.2 to CentOS 5.4 so that the virtio drivers are available. After that I was just following http://wiki.libvirt.org/page/Virtio.

For the network:

  • shut down the VM
  • edit the XML and add <model type='virtio'/> to the network section
  • start the VM
  • done

For the disk:

  • create a new ramdisk with the virtio drivers: mkinitrd --with virtio_pci --with virtio_blk -f /boot/initrd-$(uname -r).img $(uname -r)
  • or dracut -f --add-drivers "virtio_pci virtio_blk" /boot/initrd-$(uname -r).img $(uname -r) for Fedora 12
  • change /boot/grub/device.map from “(hd0) /dev/hda” to “(hd0) /dev/vda“
  • using LVM requires no changes to the root= parameter in /etc/grub.conf
  • shut down the VM
  • edit the XML changing <target dev='hda' bus='ide'/> to <target dev='vda' bus='virtio'/>
  • start the VM
  • done

During the boot of the VM I can now see that it is loading the virtio disk drivers and detecting vda1 and vda2. Using lspci and lsmod I can also verify that the new virtio devices are available and also used. The VM seems to be faster but I have not actually benchmarked it.

RPM Fusion Mirrorlist Server

Published on January 8, 2010 in fedora and mirror server. 1 Comment Tags: , , .

On the last day of the last year (2009-12-31) both RPM Fusion’s mirrorlist server were most of the time not reachable. The problem started at 00:53 (UTC) and it was at least going on until 16:00 (UTC). Both mirrorlist servers have been on the same network and the router for that network  broke down. If it would have been the link to our provider the router had a backup route to stay on-line, but this time it actually hit the single point of failure – and everything was off-line. See: error report of the provider (german).

I was never happy that both mirrorlist server were running in the same network and I especially wanted to get the mirrorlist server off my mirror server. Thanks to Patrick I have now access to another VM at a different provider where I am running a new mirrorlist server instance. It does not require much in terms of resources and bandwidth, but having root access makes everything so much easier.

RPM Fusion’s mirrorlist server are now two dedicated VMs at two different providers and that should protect the functionality from failures like the one on 2009-12-31.

Storage Trouble

Published on December 14, 2009 in mirror server. 0 Comments

In the night from Friday to Saturday a disk (slot 7) from our external RAID, containing most of the mirror server data, failed and was marked as BAD. No really a big problem, yet. The hot spare drive was activated and the rebuild started. About 24 hours later the rebuild finished. On Sunday (around 16:00) another drive (slot 5) failed and we immediately started to sync all the data to another box in case another drive decides to go off-line, which would mean a complete data loss. All the data on that RAID are (only) mirrored, but to re-sync all the 9TB we currently have would probably take a few weeks. Unfortunately the sync to another box will also take a few days until it is finished, so it is still possible that we might lose a lot. We are waiting for the replacement disks which have been promised to be here by Monday (today), but as the rebuild needs over 24 hours there is still the chance of a data loss.

Update (2009-12-14 23:20): The replacement disks have arrived and after more than twelve hours 25% of the array has been rebuilt.

Update (2009-12-15 11:00): After more than 24 hours 58% of the array has been rebuilt. It seems to rebuild faster during the night.

Back In School

Published on December 10, 2009 in adrian, grid and mirror server. 0 Comments

Not really back in school, but it has been now more than one week that I started my new job at my old university in Esslingen at the beginning of December 2009. After only 11 months at my previous workplace (Matrix Vision) I am now working for the faculty of Information Technology.

I will be responsible for the setup and installation of the new cluster of the university. The cluster will be part of the bwGRiD and it will have around 1500 cores and is currently being installed. It is partly water-cooled and a few days ago the racks were delivered and installed. The cluster is from NEC and we are expecting the servers to be delivered in the next few days. The cluster will be running Scientific Linux.

I am now in the same building as my mirror server. This might be a good thing, because now I am much closer to the hardware and can act faster if something unexpected happens… It might also be a bad thing, because now I am much closer and can experiment with things I would not do if I was not in the same building.

Lightning Talk

Published on August 19, 2009 in adrian and mirror server. 0 Comments Tags: , .

Last Thursday (2009-08-13) I gave a lightning talk about the mirror server I am maintaining. According to the description from Wikipedia lightning talks are usually between 1 and 10 minutes with a 5 minute limit being common[¹]. In this session there were four talks with each being about 15 minutes with additional 5 to 10 minutes for questions. So it was a bit longer than the definition but that was a good length for all four talks.

The lightning talks were organized by the Chaos Computer Club Stuttgart (CCCS). They are organizing a talk every month and in the summer it is usually a lightning talks session. The given talks were:

I am very happy that I decided to give my talk and altogether it was a very nice event. There was also an audio recording (which unfortunately has not yet been released).

[¹] http://en.wikipedia.org/wiki/Lightning_Talk

More Mining

Published on July 24, 2009 in mirror server. 0 Comments Tags: .

I am doing now even more data-mining in the log files of our mirror server and in the last few weeks I have added more graphical output to the information interface of our mirror server.

The first, pretty simple, new diagram is the Disk Usage By Mirrored Project.  It is created daily from a du run over the whole mirror area and then visualized using the existing code to draw a pie chart. The reason that Fedora is the largest mirrored project is due to the fact that I count all the Fedora related projects I am mirroring (fedora, archive.fedoraproject.org, secondary.fedoraproject.org) as fedora. If this diagram is now compared to the Overall Traffic Breakdown diagram it is interesting to see that there is not much difference in traffic generated by mirroring Fedora and Ubuntu, but the space required to mirror Ubuntu is much lesser. If I would know a lot more about Ubuntu I probably could start mirroring exotic Ubuntu things (just like I do with Fedora).

In addition to the new pie charts I have also created world maps on which the client distribution around the world can be seen. The code to generate the maps is based on generate-worldmap.py. With the help of this I am now able to create a world map with the client distribution for all downloads on which it can be seen that Europe is clearly the location of many (maybe even most) clients connecting to our mirror server, but also that there are basically connection coming from all over the world.

In addition to the overall map I am also creating a map for each mirrored project. It can be seen that there are mirrored projects (like Ubuntu, Fedora and OpenSUSE) which have pretty good redirectors so that only clients in the vicinity are redirected to our mirror. There are also mirrored projects (dag) which do nothing like that which results in connection from all over the world. And then there are projects which have a good redirector but do not have enough mirrors around the world like fedoraproject. What is called fedoraproject on our mirror server is a mirror of archive.fedoraproject.org (Fedora 8 and older right now, which still seems to be downloaded).

Leonidas Traffic (Part 2)

Published on June 24, 2009 in fedora and mirror server. 0 Comments Tags: .

In one of my last posts I wrote about how much Fedora related traffic we had on our mirror server during the Fedora 11 release. I got one huge comment from Jef with three questions which I am trying to answer now.

1) Assuming you could find an accurate count of EU mirrors on F10 release. Can you use the ratio of available mirrors then to available mirrors now to re-scale the activity…sort of like a mirror inflation correction to scale activity in terms of available bandwidth.

As the mirrorlist is very dynamic I do not think I can answer that.  But if somebody has some useful numbers concerning the number of EU mirrors during the F10 release as well as during the F11 release it can probably be done.

2) Can you trend the “shape” of the first week of F-11 compared to the first week “shape” of F-10 activity on your mirror?  Forget about absolute scales. Normalize each to the maximum associated with the first 24 hours of activity and see how the activity trends in time relative to that normalization. Does F-10 for example see the same second day uptick relative to the first day that you see in F-11?

That should be possible. After some gnuplot-ing I have the following diagrams:

Downloaded Data (Normalized)

This shows the data transferred for each release (normalized to the first day). It is important, however, to know that the first day of a Fedora release is not 24 hours, but only 8 hours on our mirror server as the release usually happens at 16:00 local time. Therefore I also made another diagram using the absolute numbers:

Downloaded Data (Absolute)

It can be seen that basically only the first day differs for some reason. The following days were pretty much the same, just a bit less traffic than during the Fedora 10 release. So maybe this difference is related to my assumption that there are more mirrors in Europe. Although the amount of a traffic does not differ so much the mirror server has been in a much better state during the whole release. The load used to be much higher and the http server had no free connection slots available. This time the load was not really high and after the first day it was always possible to make a http connection (although it took longer than usual).

3)I’d be really interested to know if you could identify any upticks related to F-10 downloads further away from F-10 release that correlate with ambassador activity at an EU event.

No idea, but I have the amount of data downloaded per day for each mirrored project for at least the last year available here. So if there are certain dates it can be looked up.

For my own reference these are the gnuplot commands used to create the two diagrams:

gnuplot> set terminal png size 400,300
gnuplot> set output "absolute.png"
gnuplot> set xlabel 'Days Since Release'
gnuplot> set ylabel 'Terrabytes'
gnuplot> plot 'F10' smooth csplines title 'Fedora 10', 'F11' smooth csplines title 'Fedora 11'
gnuplot> set output "normalize.png"
gnuplot> a=2.23
gnuplot> b=1.54
gnuplot> set ylabel 'Downloaded Data (Normalized)'
gnuplot> plot 'F10' using 1:($2/a) smooth csplines title 'Fedora 10', \
>'F11' using 1:($2/b) smooth csplines title 'Fedora 11'

Leonidas Traffic

Published on June 16, 2009 in fedora and mirror server. 1 Comment Tags: .

The Fedora 11 (Leonidas) release has been the Fedora release with the least pressure on our mirror server. This is probably due to the fact that there are more mirror servers in Europe than ever. In addition to the usual http/ftp/rsync traffic I had bittorrent running for the first time, but the bittorrent client was never using more than 50 MBit/s of the bandwidth (and that also dropped after the second day). Compared to the normal mirror traffic that is not really much. After running for about a week the bittorrent client has uploaded around 600 GB.

The Fedora traffic during the first few days after the release was of course higher than the average ~450 GB/day:

  • 2009-06-09: 1.54 TB
  • 2009-06-10: 2.88 TB
  • 2009-06-11: 1.62 TB
  • 2009-06-12: 1.01 TB

On the second day of the release almost 70% of the mirror traffic was Fedora related:

2009-06-10

And on the bandwidth graph it can also be seen when the bitflip was (around 15:00 local time) and when the release actually went live (16:00 local time):

2009-06-10

I do not know why the traffic dropped so significantly at around 00:00, but probably our mirror was dropped from the mirror list, because the crawler (from MirrorManager) was no longer able to connect and verify that our mirror was up to date.

Following Fedora 11

Published on June 10, 2009 in fedora and mirror server. 0 Comments

Syncing Fedora 11 to my mirror went pretty well. With hardlinks it took not much longer than 24 hours to transfer all the data.

On Monday rawhide was updated again and my mirror scripts started. This time it took more than 55 hours for the first run and it just finished a few minutes ago. I re-started the mirror script immediately after it finished because in the 55 hours it was running rawhide had already changed again. So it looks like to get my rawhide mirror in sync with the master server it will take more than three days. The slowness is probably not only caused by the large number of changes but also by the fact that it is now also necessary to transmit all the DeltaRPMS. Somehow it should be possible to use the DeltaRPMS for mirror syncing. Another reason might be that my mirror server has been pushing Fedora 11 data out at about 1Gbit/s, which probably does not help to speed things up.