Server trouble

Matt Mackall mpm at selenic.com
Fri Mar 1 02:20:37 UTC 2013


Up until Sunday the main mercurial server had been running without
trouble for 549 days. Then it encountered an i/o error on the xen root
partition that could only be addressed with a reboot. I took advantage
of the downtime to do a long overdue kernel upgrade.

Unfortunately, the new kernel only lasted a few days before deciding
to randomly disable the primary network interface twice in 24 hours.
Worse yet, this coincided with me leaving the country on a two week
vacation. So it took most of a day to get someone on-site and on the
phone with me to fix things.

I'm pretty sure that I've now got a cron job that will detect and fix
the problem until I can do a proper fix, but we might be in for a
rough couple of weeks.

-- 
Mathematics is the supreme nostalgia of our time.


More information about the Mercurial-devel mailing list