This is not of huge concern, but twice in the past few months I have had fail2ban die with a weird error. These are the logs starting from an hour before the problem:

2009-01-30 14:58:07,077 fail2ban.actions: WARNING [ssh-iptables] Ban 157.100.18.3
2009-01-30 15:23:32,250 fail2ban.actions: WARNING [ssh-iptables] Ban 202.97.194.169
2009-01-30 15:23:32,258 fail2ban        : ERROR  (514, 'Unknown error 514')
Traceback (most recent call last):
  File "/usr/bin/fail2ban-server", line 126, in start
    self.__server.start(self.__conf["socket"], self.__conf["force"])
  File "/usr/share/fail2ban/server/server.py", line 87, in start
    self.__asyncServer.start(sock, force)
  File "/usr/share/fail2ban/server/asyncserver.py", line 135, in start
    asyncore.loop()
  File "/usr/lib/python2.4/asyncore.py", line 192, in loop
    poll_fun(timeout, map)
  File "/usr/lib/python2.4/asyncore.py", line 122, in poll
    r, w, e = select.select(r, w, e, timeout)
error: (514, 'Unknown error 514')
2009-01-30 15:23:34,261 fail2ban.actions: WARNING [ssh-iptables] Unban 157.100.18.3
2009-01-30 15:23:34,272 fail2ban.actions: WARNING [ssh-iptables] Unban 202.97.194.169

[snip to 1 hour before next crash]

2009-04-28 11:07:18,124 fail2ban.actions: WARNING [ssh-iptables] Ban 61.54.243.202
2009-04-28 11:11:27,128 fail2ban.actions: WARNING [ssh-iptables] Ban 82.207.66.14
2009-04-28 11:22:39,051 fail2ban.actions: WARNING [ssh-iptables] Unban 124.30.74.62
2009-04-28 11:37:18,238 fail2ban.actions: WARNING [ssh-iptables] Ban 124.30.74.62
2009-04-28 11:50:44,316 fail2ban.actions: WARNING [ssh-iptables] Ban 74.13.225.76
2009-04-28 11:50:44,356 fail2ban        : ERROR  (514, 'Unknown error 514')
Traceback (most recent call last):
  File "/usr/bin/fail2ban-server", line 126, in start
    self.__server.start(self.__conf["socket"], self.__conf["force"])
  File "/usr/share/fail2ban/server/server.py", line 87, in start
    self.__asyncServer.start(sock, force)
  File "/usr/share/fail2ban/server/asyncserver.py", line 135, in start
    asyncore.loop()
  File "/usr/lib/python2.4/asyncore.py", line 192, in loop
    poll_fun(timeout, map)
  File "/usr/lib/python2.4/asyncore.py", line 122, in poll
    r, w, e = select.select(r, w, e, timeout)
error: (514, 'Unknown error 514')
2009-04-28 11:50:45,443 fail2ban.actions: WARNING [ssh-iptables] Unban 61.54.243.202
2009-04-28 11:50:45,457 fail2ban.actions: WARNING [ssh-iptables] Unban 82.207.66.14
2009-04-28 11:50:45,480 fail2ban.actions: WARNING [ssh-iptables] Unban 124.30.74.62
2009-04-28 11:50:45,499 fail2ban.actions: WARNING [ssh-iptables] Unban 74.13.225.76

Looks like fail2ban is nice enough to clean up after itself (unbans any IPs currently banned), but it doesn't ban any new addresses until it is restarted. The Linode is running the "latest" kernel - on 30 Jan this was 2.6.18.8-linode10 and on 28 Apr was 2.6.18.8-linode16 (both on Xen). Some searching around for this error found an lkml thread from early 2007 and related blog entry which seemed to indicate this is not a Python bug, but may lie in the kernel. A RHEL bug report hints at a late 2007 fix for a similar (?) kernel problem, but I don't know if their patch made it upstream.

Perhaps if someone gets bored they could check into whether the Red Hat patch made it into mainline and, by implication, Linode's build. Bonus points for figuring out whether that patch actually fixes the problem. :-)

Hmm, looks like I should have checked fail2ban's site - apparently they have a new version that "fixes" the problem by using poll() instead of select(). Not sure if that means there's still something odd lurking in the kernel somewhere...