ping timeout is in milliseconds, not seconds

[Brian Landy]

Hi, when defining a test like this:

if failed ping4 count 3 with timeout 5 seconds then alert

The units are actually milliseconds, not seconds. If I change the test to read "5000 seconds" and use an offline IP, I can time the interval between ping failed messages written to the log and see that they come every 5 seconds. I verified the issue on 5.18, 5.20, and a pull from git. I'm running on OpenBSD 6.0.

Thanks, Brian

[Tildeslash]

Hi,

i'm unable to reproduce the issue using development version on OpenBSD 6.0 amd64 ... added unix timestamp to the debug message, it shows there are 5 seconds between attempts:

1485457391: Ping response for 192.168.50.250 1/3 timed out -- no response within 5 s
1485457396: Ping response for 192.168.50.250 2/3 timed out -- no response within 5 s
1485457401: Ping response for 192.168.50.250 3/3 timed out -- no response within 5 s
'nohost' ping test failed
'nohost' icmp ping failed, skipping any port connection tests

Monit configuration:

set daemon 5

set httpd port 2812 allow localhost

check host nohost with address 10.2.3.250
        if failed ping4 count 3 with timeout 5 seconds then alert

There was one ping related issue (#488) fixed in the development version, please can you test it?:

wget https://bitbucket.org/tildeslash/monit/get/master.tar.gz

To compile:

./bootstrap
./configure
make

[Brian Landy]

I'll test that shortly and get back to you.

Interesting, here's what I am seeing with "5000 seconds":

 [EST Jan 26 15:23:23] error    : Ping response for 192.168.170.111 1/3 timed out -- no response within 83.333 m
 [EST Jan 26 15:23:29] error    : Ping response for 192.168.170.111 2/3 timed out -- no response within 83.333 m
 [EST Jan 26 15:23:35] error    : Ping response for 192.168.170.111 3/3 timed out -- no response within 83.333 m

monitrc:

check host 192.168.170.111 with address 192.168.170.111
   if failed ping4 count 3 with timeout 5000 seconds 1 cycles then alert

[Brian Landy]

Even on this version I'm seeing the shorter times:

 check host 192.168.170.111 with address 192.168.170.111
   if failed ping4 count 3 with timeout 5 seconds 1 cycles then alert

 [EST Jan 26 17:23:27] info     : 'firefly.landy.cx' Monit 5.21.0 started
 [EST Jan 26 17:23:28] error    : Ping response for 192.168.170.111 1/3 timed out -- no response within 5 s
 [EST Jan 26 17:23:29] error    : Ping response for 192.168.170.111 2/3 timed out -- no response within 5 s
 [EST Jan 26 17:23:30] error    : Ping response for 192.168.170.111 3/3 timed out -- no response within 5 s

 check host 192.168.170.111 with address 192.168.170.111
   if failed ping4 count 3 with timeout 5000 seconds 1 cycles then alert

 [EST Jan 26 17:24:41] info     : 'firefly.landy.cx' Monit 5.21.0 started
 [EST Jan 26 17:24:47] error    : Ping response for 192.168.170.111 1/3 timed out -- no response within 83.333 m
 [EST Jan 26 17:24:53] error    : Ping response for 192.168.170.111 2/3 timed out -- no response within 83.333 m
 [EST Jan 26 17:24:58] error    : Ping response for 192.168.170.111 3/3 timed out -- no response within 83.333 m
 [EST Jan 26 17:24:58] error    : '192.168.170.111' ping test failed

Brian

[Tildeslash]

Fixed: Issue #543: Ping test may timeout prematurely on machine with heavy ICMP traffic.

→ <<cset 055bb736431b>>

[Tildeslash]

It should be fixed now ... please can you fetch the development snapshot again? (wget https://bitbucket.org/tildeslash/monit/get/master.tar.gz)

The problem was in timeout handling, where delta in microseconds was subtracted from timeout in milliseconds.

[Brian Landy]

Yes, that fixed it for me. Thank you!