Re: monit with nginx

[Derek]

The access log shows this

Monit 4.10.1
ip.removed - - [26/Oct/2010:23:46:58 +1300] "GET / HTTP/1.1" 400 173 "-" "-"

Monit 5.0.3
ip.removed - - [26/Oct/2010:23:47:54 +1300] "GET / HTTP/1.1" 200 151 "-" "monit/5.0.3"

Just ngrep'ed the GET requests to nginx.

Monit 4.10.1
GET / HTTP/1.1..Host: ..Accept: */*..Connection: close..User-Agent: monit/4.10.1....

Monit 5.0.3
GET / HTTP/1.1..Host: ip.address.here..Accept: */*..Connection: close..User-Agent: monit/5.0.3....

It looks like the IP/domain is left from the Host: area of the GET request on monit 4.10.1

I'm not that much of a http protocol guru - does this help find the cause.

Not lenny-backports packages for monit on debian.. might have to compile up a new version for my servers,

Thanks
Derek

On 26/10/10 11:00 PM, António P. P. Almeida wrote:

On 26 Out 2010 10h33 WEST, address@hidden wrote:

[1  <text/plain; ISO-8859-1 (8bit)>]
Thanks for the reply,

I'm using this monit instance to check other nginx servers for 
availability, so this is done all remotely using public IP's.

You can also use the hostname. 

There are local checks running on each web server to monitor nginx
locally also, this was using a tcp check on port 80.  However I
updated the local checks to your line below (if failed port 80
protocol http then restart).  This also gave the 400 back using
localhost..

Hmmm. 400 is bad request. What do you see in Nginx logs? 

In both cases the syslog shows..

HTTP error: Server returned status 400

'domain.co.nz' failed protocol test [HTTP] at
INET[123.456.789.10:80] via TCP

What version of monit are you running?

I'm running monit 5.2.1 and nginx 0.8.53. 

--- appa