[tor-relays] Tor crashes frequently on fast relay

[Stephan]

On 11.09.2013 13:33, Random Tor Node Operator wrote:
> Could you post the corresponding line(s) in your monitrc?

Of course. I use the default of "set daemon 120", so tor is checked once 
every 120 seconds ('one cycle'). The tor specific part of the 
configuration is this:

    if failed host 127.0.0.1 port 9030 type tcp for 10 cycles then 
restart
    if failed host 127.0.0.1 port 9001 type tcp for 10 cycles then 
restart
    if 5 restarts within 100 cycles then timeout

I'm still experimenting with the exact numbers. Restarting after only 
one failed check was obviously too fast. Waiting for 10 consecutive 
failed checks (i.e. tor is not responding for 20 minutes) may be too 
long, but I don't want to ruin the 'stable' flag only because a botnet 
is wreaking havoc for a few minutes.

The last line is kind of a safeguard. I get monit alerts by email. If 
something is seriously broken with my server (or tor) five mails will 
suffice. Even if I am on vacation for a few weeks, I don't need a 
reminder every 20 minutes. ;-)


-Stephan