[racket] recent server outage

From: Robby Findler (robby at eecs.northwestern.edu)
Date: Wed Jan 26 08:12:32 EST 2011

We recently had an outage of a bunch of Racket-related web services
(notably planet).  Our apologies.  Things should be back to normal,
but if you spot any problems, please let us know.

The longer version of the story: Eli detected a possible break-in on
one of our servers; someone or something had replaced ssh with a
binary that didn't have the executable bit set.  So, he immediately
brought down the machine.

We did not find any other modifications to any files (the files served
by planet, in particular, had not been changed from a recent backup).
To prevent further problems, the machine was upgraded to the most
recent Fedora and is now using selinux.  Most of the downtime was
dealing with little changes compared to the old version of the OS and
sorting out selinux permissions.  Hopefully this will help protect us
more in the short term.  We're also looking into other, longer term
solutions; we'll keep you posted as we sort that out.

Eli, Robby, Matthew, Jay, Matthias


Posted on the users mailing list.