Why do we drink cow's milk? Who was the first guy who first looked at a cow and said "I think I'll drink whatever comes out of these things when I squeeze 'em!"? -- Calvin

ftp-stud.hs-esslingen.de
primary server load:9.10 9.41 11.09
secondary server load:
primary server http connections:12096 (IPv6 701)
secondary server http connections:

Current bandwidth utilization 596.41 Mbit/s (primary server)
Bandwidth utilization bar
Current bandwidth utilization 0.00 Mbit/s (secondary server)
Bandwidth utilization bar

56796.32 TB transmitted since 2007-12-20
On average 9.79 TB per day
with a peak of 52.18 TB on 2024-03-28
47.50 TB transmitted yesterday
2.19 TB transmitted today


This is the mirror server cluster of the University of Applied Sciences Esslingen. Its 10 Gigabit connections feed directly to the university backbone, which is connected with 20000 Mbit to the Internet.

News:


2022-01-09: Upgraded to Fedora 35
2021-01-30: Upgraded to Fedora 33.
2020-01-11: Upgraded to Fedora 31.
2018-12-30: Upgraded to Fedora 29 and finally enabled SELinux.
2018-01-08: Upgraded to Fedora 27
2018-01-05: Moved to new hardware. 256GB RAM. 16 cores (32 with hyperthreading)
2017-02-23: Added IPv6 reverse record for 2001:7c0:700::10 - rhlx01.hs-esslingen.de
2017-02-21: Added AAAA record for rhlx01.hs-esslingen.de - 2001:7c0:700::10
2017-02-20: Enabled IPv6 address 2001:7c0:700::10
2016-10-23: Enabled re-direct to HTTPS for http://lisas.de/ and http://ftp-stud.hs-esslingen.de/info/.
2013-08-08: Updated to Fedora 19 (Schrödinger’s Cat)
2012-11-22: New mirror added: FreeBSD
2012-11-15: Installed 10TB of additional disk space. There are now roughly 30TB of disk space used for mirroring.
2012-10-04: Moved to new hardware. 128GB RAM. 12 cores (24 with hyperthreading).
2011-06-25: Over 10TB on one day. All thanks to Firefox: 10.14TB
2011-06-23: Again a traffic record. Almost 10TB on one day.
2011-04-30: Yesterday we established a new traffic record. 8.13TB. Thanks to the combined release of Ubuntu 11.04 and Firefox.
2011-04-01: All of a sudden it is a cluster! For over ten years a single machine was serving all the files we are mirroring. Today we started to use a second machine to distribute the load hoping to improve the user experience.
2011-02-17: Added new mirror: Mageia.
2010-09-28: Added new mirror: The Document Foundation.
2010-09-15: Added new mirror: mariadb.
2010-02-26: This server will not be reachable tomorrow (Saturday 2010-02-27) due to campus wide network changes for at least 8 hours.
2010-02-04: Added new mirror: Scientific Linux.
2009-12-22: After mirroring ftp.mozilla.org since 2000 we are now finally included in the automatic redirector pool. Chances are, if you are downloading something from mozilla.org, that you will be redirected to our mirror.
2009-11-10: We are now part of the rsync.de.gentoo.org and rsync.europe.gentoo.org rotation.
2009-10-09: Reboot for RAID reconfiguration and kernel upgrade.
2009-07-05: Added new mirror: neooffice.org.
2009-02-14: $ date +%s
1234567890
2009-01-11: The PHP scripts behind our information service are now available through: git clone http://ftp-stud.hs-esslingen.de/info/git/info/
2009-01-01: Jan 1 00:59:59 rhlx01 kernel: Clock: inserting leap second 23:59:60 UTC
2008-11-17: In the last four weeks the data has been moved to a new, even bigger RAID. The amount of RAM has also been increased. The system is now running with 40GB of RAM and over 10TB of diskspace.
2008-11-16: Added new mirror: ELDK.
2008-07-07: Started to mirror fedora-secondary.
2008-07-01: Short unannounced downtime to boot a new kernel.
2008-04-17: Downtime for a few hours to install a fast (15K) 300GB SAS drive. This disk is used to act as a disk cache for apache. The goal is to reduce the number of simultaneous accesses on our main RAID. We also changed the bonding mode to: mode=6 (balance-alb)
2008-04-09: Stopped the fedora and fedora.extras mirror scripts. These scripts were pointing to rsync modules which were removed on the master server because they were only necessary up to Fedora 6 which is already EOL since a few months.
2008-02-16: Today was for a few hours one of the RAIDs offline to do the following:

tune2fs /dev/sdb1 -O dir_index

e2fsck -C 0 -f -D /dev/sdb1

This should, if correctly informed, speed up lookups in large directories.
2008-02-12: Rebooted for a new kernel to fix the current available local root exploit.
2008-01-21: New mirror of http://www.debian-multimedia.org/ at http://ftp-stud.hs-esslingen.de/pub/Mirrors/debian-multimedia/.
2008-01-15: All volumes are mounted again. Back at normal operation.
2008-01-14: We unmounted /ftp/pub/.2 again because of problems with the backend. This means that following mirrors are currently not available: ccux-linux.de, ftp.kde.org, ftp.rootlinux.org, ftp.xfree86.org, ftp.ximian.com, Mandrivalinux and wikipedia. Estimates are that it will be fixed sometime tomorrow.
2008-01-11: Thanks to Patrick's mail I finally fixed a SQL injection security bug in our mirror status page. I was actually able to create and drop tables, but it seems nobody looks much at that page because it is at least online for two years and it has never been compromised (as far as we know). The php code was actually fixed with Gunnar's help.
2007-11-09: Disabled debian-non-US mirroring, because it is obsolete since Debian 3.1, which was in 2005.
2007-09-23: After trying out different new Fedora 7 kernels (which led to a few crashes and downtime) we switched to a CentOS kernel: 2.6.18-8.1.8.el5
2007-07-06: On of our volumes (/ftp/pub/.2) was since 6 in the morning not available due to changes in the server room. Before going back online it needed unfortunately a fsck which took quite some time. But now we are fully operational again.
2007-05-28: Unscheduled downtime due to building wide power outage.
2007-05-07: After a backup of all our FTP data the RAID will be deleted and a new filesystem will be created. We are going through all this effort because the fragmentation on our RAID has become so bad that the performance suffers immensely. As soon as the new filesystem is ready we will start the restore process to get all our data back.
2007-01-06: Stopped mirroring download.fedoralegacy.org as the project has stopped providing any new updates. The data will stay available on this mirror.
2006-12-18: Two filesystem related crashed since we moved to the new hardware. It seems like the x86_64 port of Linux is not as well tested as the i386 port. We are thinking about moving to i386 again which would be unfortunate but if it is necessary we will do it. Right now we are still watching how the system behaves.
2006-12-14: The new server arrived yesterday and since today we are running from the new 64bit hardware.
2006-11-17: Changed the Mandriva mirror script to be two scripts: one that mirrors the official tree and the other mirrors the devel tree. The Mandriva mirror has become very big and until now the script normally never finished before the next instance was started.
2006-11-14: Attached a new RAID (StorEdge 3310 from SUN). It is not really new but it has 671GB and our big 4TB RAID is reaching its limits.
2006-10-31: Downtime today was a bit longer than expected. But we finally made some long outstanding changes. We are now using grub instead of lilo and with this change we enabled the serial console in the BIOS. So we now have the possibility to control the output from power-on until the OS is running. We also finally enabled the bonding of the two network cards we have so that we are now able to have an maximum output of 2GBit/s.
2006-10-23: Updated to Fedora Core 6. A reboot is still required to get the newest kernel running.
2006-04-21: Many reboots to get a newer kernel running but all attempts failed.
2006-04-02: Glory and fame for this server: FTP-Server mit Massendurchsatz.
2006-03-31: Reboot to install a second network adapter and to boot a newer kernel.
2006-03-19: Upgrade to Fedora Core 5.
2006-01-19: After a reboot the /ftp partition was not mounted anymore. According to dmesg: qla2300 0000:03:05.0: Cable is unplugged.... This should be fixed pretty soon.
2005-10-29: Yesterday a router has been replaced which caused an outage of about 30 minutes. The good thing about this outage was that the shaping of our bandwidth usage has been removed and it immediately jumped to over 150 Mbit/s. So now were back in the real business.
2005-10-13: Due to errors on the uplink between Esslingen and Stuttgart a bandwidth shaping (80 Mbit/s) has been activated by our provider.
2005-08-22: New Mirror added: CentOS
2005-08-14: New Mirror added: openSUSE
2005-07-15: A new kernel was installed and a very big RAID (4TB). The new RAID is not in use yet because were are still doing tests on how to use it. The RAID is attached via Fibre Channel and consists of 12 400GB SATA drives.
2005-06-13: Since Friday I was updating the server to Fedora Core 4 and since today it is serving Fedora Core 4 and running Fedora Core 4.
2005-04-21: After about 150 days without a reboot a hardware re-configuration made a reboot necessary. Two RAIDs were removed and one was added and now we have 100GB additional space which of course will also be used for the FTP mirror. The two RAIDs were over 7 years old and together they only provided around 230GB. The new RAID is also old hardware but probably only 4 years old and it provides 350GB. In addition to the new hardware there was also a new kernel installed before the reboot.
2004-12-02: The whole university was disconnected from the Internet due to construction work during which our uplink was cut. It took them some hours to fix it but now everyting is back to normal. Outage was from 7:30 to 12:25.
2004-11-26: New Mirror added: releases.ubuntu.com
2004-11-24: Reboot to finish the upgrade to Fedora Core 3 and to install a new network card. After many years our network connection is again using copper. We are also using a different driver for our internal RAID controller. Instead of the dpt_i2o module we are now using the i2o_block module. Uptime was 85 days.
2004-11-19: New Mirror added: aurox
2004-10-30: Some process(es) consumed all the memory and the machine was not usable for about an hour. This was the time the OOM killer striked and started killing about half of the processes. After restarting all services everything was back to normal without rebooting.
2004-10-07: New Mirror added: impi
2004-09-20: New Mirror added: ccux-linux
2004-08-31: Crash. Reboot.
2004-08-27: Reboot.
2004-07-09: Reboot... But the reason will not be told.
2004-06-18: Our shiny new 2.6.7 kernel crashed. Recompiled with swap enabled and two reboots.
2004-06-16: Many reboots and finally a 2.6.7 kernel is up and running.
2004-06-09: Reboot to install a different network card. The 3COM Gigabit card has been replaced with an Intel e1000.
2004-06-01: Two new mirrors available: gentoo and fedoralegacy.org
2004-05-28: Downtime was one hour today. This was necessary to make hardware changes. Old RAID removed from rack. New RAID installed into the rack. New memory installed in the RAIDs.
2004-05-16: The 180GB RAID was upgraded to 600GB. The kernel was updated to support XFS and more than 4GB RAM. Fedora Core 2 was installed two days ago.
2004-05-14: A new RAID with additional 600GB was attached. New kernel installed.
2004-03-17: The breakdown of an UPS powering three of our external RAID arrays required a reboot of the system.
2004-01-08: Once again a reboot to update the kernel to 2.4.24.
2003-11-17: One reboot was necessary to connect on new RAID and to change the layout and configuration of one of the already connected RAIDs. The server has now 903GB of diskspace which is used for the mirrors.
2003-11-02: Our provider disabled our whole network for the last two days to investigate network errors on our uplink. We used this network downtime to run a fsck over all our big RAIDs and could therefore make something useful during the two days nobody was able to connect to us. We were also able to do some minor repartitioning and are now running with Linux 2.4.22-ac4.
2003-10-15: The instabilities of the last weeks seem finally over. A new SCA backplane was installed, all RAIDs are working as expected and no more SCSI errors are reported. So we are prepared to create new uptime record for this server.
2003-10-07: In the last weeks there were many unannounced reboots to update the bios of every component because one of our RAIDs is refusing to use all available disks and this is the only help the manufacturer is offering us.
2003-10-01: A new mirror has been added: auroralinux.org
2003-09-17: 2.4.22 didn't work very good and therefore we are using again 2.4.21-rc2-ac2 with 64GB support.
2003-09-16: Downtime to upgrade the hardware. Two additional CPUs. 4GB of additional RAM and 150GB new harddisk space. A new kernel has also been installed (2.4.22).
2003-07-21: The ftp and http processes were disabled for a short time to unplug the external RAIDs and connect them to a new UPS.
2003-05-15: Reboot to install new kernel: 2.4.21-pre2-ac2
2003-05-14: Big change: The IP address has been changed from 134.108.34.10 to 129.143.116.10. Right now it seems to work quite normal but we will see how this change will influence the current status. More information here when the transition is finished.
2003-04-28: The SuSE mirror has been completley redesigned. It is now mirrored from a more official place and the directory structure makes therefore now more sense. Get SuSE here.
2003-04-07: RedHat 9 has been released. Also available via apt.
2003-04-04: A new mirror has been added. The eclipse project is now also available from here. Get eclipse.
2003-04-02: This server has been updated to RedHat 9.
2003-03-28: New record. Yesterday the total outgoing traffic was over 11TB. Never before has ftp-stud.fht-esslingen.de delivered so much data.
2003-03-25: Mandrake 9.1 has been released which resulted in much more bandwidth consumption. Instead of the usual 30-50MBit/s it is now always more than 200MBit/s and all 500 available ftp connections are in use.
2003-03-20: Reboot was necessary to patch the kernel (ptrace exploit fix)
2003-03-06: Just a status report, that the new kernel is operating without any apparent problems.
2003-03-05: Right now we are running again without any bandwidth shaping which immediately resulted in a tripled bandwidth consumption. This server is the machine which produces the highest amount of traffic in the BelWü. Should this make us proud?
2003-03-04: Short, unannounced, downtime to upgrade the kernel. Now a pure and selfmade 2.4.20 is powering this server.
2003-02-28: Maintenance reboot to install two new SCSI controllers. This leads to that this server is now independent of any other server and no data is mounted via NFS but locally available. Two hours later after trying to unmount the last NFS-mounted file-system the system froze again completely. Another reboot and time to install a new and better kernel.
2003-02-19: The new RedHat beta (phoebe) has been released. This is the third version of this RedHat beta release. The apt repository has also been updated.
2003-02-18: Two crashes today. The kernel halted after an unmount of loopback mounted iso image.
2003-01-15: The apt repository from rawhide seems now fully functional.
2003-01-14: debian-security is now mirrored from security.debian.org and debian from ftp.de.debian.org. Both were previously mirrored from ftp.rfc822.org.
2003-01-07: The mirror of ftp.gimp.org is now updated again.
2002-12-24: The RPMs mirrored from redhat are availabe via apt-get.
2002-12-23: The new Red Hat beta "PHOEBE" is available.
2002-12-22: New mirror added: knoppix
2002-12-21: New mirror added: freshrpms.net
2002-12-20: Public rsync access available
2002-12-19: New Hardware

hand crafted with vim [ Powered by Fedora ] Valid CSS! Valid XHTML 1.0!

Page creation time: Fri, 19 Apr 2024 22:59:55 GMT