"Michael O'Reilly" writes:
>
>Doing a while sleep 1 ; do echoping -h ... ; done
>
>Elapsed time: 3.209105 seconds
>Elapsed time: 0.043567 seconds
>Elapsed time: 3.253924 seconds
>Elapsed time: 0.063965 seconds
>Elapsed time: 0.037700 seconds
>Elapsed time: 0.039206 seconds
>Elapsed time: 0.029647 seconds
>Elapsed time: 0.031966 seconds
>Elapsed time: 3.208050 seconds
>Elapsed time: 0.042787 seconds
>Elapsed time: 0.073833 seconds
>Elapsed time: 0.051523 seconds
>Elapsed time: 3.218910 seconds
>Elapsed time: 0.036277 seconds
>
>This is to a busy cache, on the same 100meg full duplex ethernet,
>which zero packet loss between the two machines. The echoping is
>running every second, so the object is hot in the cache. The object is
>a normal html page, no cgi or nothing. The access logs show a
>cache_hit on each access.
>
>Why are the 3 second response times coming up? They keep on appearing
>at irregular intervals.
>
>Anyone think of a something that would produce a 3 second delay on a
>transaction??
The v1.2 client program can be used to make the same test, and you
don't need to use the shell to sleep and fork for you:
surf-wessels ~ 2> client -g 0 -I 1 http://www.nlanr.net/
1998-02-17 10:25:38 [1]: 0.236 secs
1998-02-17 10:25:39 [2]: 0.009 secs
1998-02-17 10:25:40 [3]: 0.009 secs
1998-02-17 10:25:41 [4]: 0.011 secs
1998-02-17 10:25:42 [5]: 0.010 secs
1998-02-17 10:25:43 [6]: 0.010 secs
1998-02-17 10:25:44 [7]: 0.009 secs
1998-02-17 10:25:45 [8]: 0.010 secs
Duane W.
Received on Tue Jul 29 2003 - 13:15:46 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:11:43 MST