Reboot Schedule

Sun, Feb 13th, 2005 - 3:00PM EST / 8:00PM UTC
Host 1, 2, 6, 13, 15, 20, 26, 28, 29, 31

Reboot Details

All Linodes will be shutdown properly, the host will be rebooted, and then your Linode will be automatically rebooted using the last used config profile. You do not need to issue reboots. Actual down-time for each Linode should be between 15-30 minutes.

Please allow a four hour window for the reboot of your host to occur. I'll be doing each host one at a time, in the order they appear above.

Bug Fix Matrix

host1: NET_FIX, SYSEMU, CFQ-TS
host2: NET_FIX, RAID_RESYNC, SYSEMU, CFQ-TS
host6: SIGNALMASK_FIX, RAID_RESYNC, CFQ-TS
host13: RAID_RESYNC, SYSEMU, CFQ-TS
host15: RAID_RESYNC, SYSEMU, CFQ-TS
host20: SIGNALMASK_FIX, RAID_RESYNC, CFQ-TS
host26: SIGNALMASK_FIX, CFQ-TS
host28: SIGNALMASK_FIX, RAID_RESYNC, CFQ-TS
host29: SIGNALMASK_FIX, RAID_RESYNC, CFQ-TS
host31: SIGNALMASK_FIX, CFQ-TS

Legend

NET_FIX
Fixes a bug in the e100 driver which occasionally caused these machines (and the Linodes on them) to drop off the network for a short while.

RAID_RESYNC
Fixes a bug in the host kernel that prevents a removed device from being re-added back into the RAID array.

SIGNALMASK_FIX
Fixes a bug in SKAS introduced in the 2.6.7 host kernel, which prevents strace from working correctly and also caused the mysql instabilities.
http://www.linode.com/forums/viewtopic.php?t=1017

SYSEMU
Provides the SYSEMU performance patches where these hosts previously didn't not. For example, roughly a 33% performance increase with mysql.
http://www.linode.com/forums/viewtopic.php?t=902

CFQ-TS
Completely Fair Queuing, Time-Sliced disk I/O scheduler. Provides improved disk throughput and fairness under contention.
http://www.linode.com/forums/viewtopic.php?t=1330


If this reboot affects you, you'll be getting an email from me shortly.

Status and details of each reboot will be posted on this thread. You can also follow along in the Linode chat room (#linode on oftc).

-Chris

Could you try and keep host13's downtime to 1 minute? That's the only way I can get 99.999% uptime for the past 130 odd days.

On a more serious note, I wish in the future you would give more time between the announcement and the reboot.

On the hand, thanks for keeping your systems up to date. I'm excited about the kernel upgrades and my Linode has needed its kernel updated for a long long time.

Anyone care to quick scramble and gather some performance statistics before the upgrade for comparison?

pants

I think that before and after stats might be very interesting. I'm on host 9 though, so it won't be me.

_________________
/ Peter

[root@host1 root]# uptime
15:01:06 up 417 days, 20:26, 4 users, load average: 2.94, 2.48, 2.59

# uname -a
Linux host1.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host1 upgrade complete

[root@host2 root]# uptime
15:18:18 up 417 days, 18:43, 4 users, load average: 2.98, 1.85, 1.76

[root@host2 root]# uname -a
Linux host2 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host2 upgrade complete

[root@host6 root]# uptime
15:34:40 up 240 days, 17:31, 4 users, load average: 3.65, 3.32, 3.12

[root@host6 root]# uname -a
Linux host6.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host6 upgrade complete.

[root@host13 root]# uname -a
Linux host13.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host13 upgrade complete.

[root@host15 root]# uname -a
Linux host15.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host15 upgrade complete.

[root@host26 root]# uptime
16:41:56 up 226 days, 18:03, 2 users, load average: 2.86, 3.48, 4.25

[root@host26 root]# uname -a
Linux host26.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU

Host26 reboot complete.

[root@host20 root]# uptime
16:39:59 up 172 days, 21:14, 2 users, load average: 2.92, 2.37, 2.33

[root@host20 root]# uname -a
Linux host20.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host20 upgrade complete.

[root@host28 root]# uname -a
Linux host28.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host28 upgrade complete.

[root@host29 root]# uptime
17:25:21 up 163 days, 22:52, 3 users, load average: 2.54, 1.75, 1.46

[root@host29 root]# uname -a
Linux host29.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host29 upgrade complete.

[root@host31 root]# uptime
17:37:09 up 163 days, 23:04, 475 users, load average: 3.82, 3.37, 3.05

[root@host31 root]# uname -a
Linux host31.linode.com 2.6.10-3-bigmem #1 SMP Sun Feb 13 13:49:42 EST 2005 i686 i686 i386 GNU/Linux

Host31 upgrade complete.


This concludes our scheduled upgrades. Thanks for tuning in.

-Chris

I must say I'm almost disappointed at how uneventful the upgrades were! I started monitoring the progress a bit late, and my Linode had been running fine for over 30 minutes!

Thanks for the smooth upgrades and hard work Chris. The only problem/outtage I've experienced at Linode has been some severe network congestion at one of the data centers ages ago.