[sf-lug] done: sf-lug.com to have short reboot (was down for about 12.5 minutes late yesterday evening)

Michael Paoli Michael.Paoli at cal.berkeley.edu
Tue Jan 22 07:13:02 PST 2013

Well, was a moderate bit longer than anticipated.

sf-lug.com. (webserver) was down for about 12.5 minutes[1]:
up 2013-01-22 06:16:28+00:00 (2013-01-21 22:16:28-08:00)
up 2013-01-22 06:28:58+00:00 (2013-01-21 22:28:58-08:00)

Turns out on the reverse dependency host (vicki), there was an
/etc/fstab entry that hadn't been properly updated.  So, upon reboot,
fsck, somewhat misleadingly, complained about "corrupt" filesystem and
wanted manual intervention to fix it.  Once the actual (/etc/fstab)
error was found and corrected (blkid came in quite handy), all appeared
fine, and one additional reboot of vicki was performed to help ensure
all came up fine when (re)booting.  There was also a bit of additional
delay from fsck filesystem checks on ext3 filesystems having been past
their configured maximum days and/or number of mounts since last
checked, but those went relatively quickly (took maybe a few minutes or
so).  Good to reboot host at least once in a great while - to ensure
all obsoleted files (e.g. older versions of files prior to security and
bug fix updates) are released (no longer open) and no longer used, and
that all the boot initialization (e.g. startup scripts/programs) work
as expected (and configuration errors or bugs haven't been accidentally
introduced).  Host vicki hadn't been rebooted since, I believe
2012-02-27[2] - so it had been a rather long while.

$ expand -t2 < ck_sf-lug.com.
IP= # sf-lug.com.
stopfile=.$(basename $0).stop
while :
   nc -w "$timeout" -z "$IP" 80 && {
     echo -n 'up '
     TZ=GMT0 date +'%Y-%m-%d %H:%M:%S+00:00'
   sleep "$sleep"
   [ -f "$stopfile" ] && {
     rm "$stopfile"


> From: "Michael Paoli" <Michael.Paoli at cal.berkeley.edu>
> Subject: sf-lug.com to have short reboot (should be well under 10  
> minutes - for maintenance)
> Date: Mon, 21 Jan 2013 22:10:52 -0800

> The sf-lug.com virtual host (and it's reverse dependency vicki and its
> dependencies) will be doing a maintenance reboot.  Anticipated this
> will take (well) under 10 minutes.  Be back shortly.  I'll update
> later, but that may be fair while after the reboot.

More information about the sf-lug mailing list