[EMAIL PROTECTED] wrote:
The problem we are up against is that when we look at a network trace on
the view host- we see packets being responded to after hundreds or
thousands of milliseconds by our view_server process. But, when we look
at the system call trace for the view_server, it is responding to RPCs
very crisply. So there is clearly a queue forming in the incoming IP and
UDP layers. The questions we would have are:
1. how can we measure the size of the queue in the incoming UDP
socket for the view_server process?
BTW, as someone here (thanks Ey!) pointed out, the above question may be more
easily answered by using mdb:
You can find the address of the conn_t (first column in output) by doing
'ndd -get /dev/ip ip ip_conn_status'. Then if you do the following
with the <conn_t addr> for the connection of interest:
# mdb -k
mdb> <conn_t addr> ::print conn_t conn_sqp->sq_count
you should get an idea of the queue size.
hope that helps.
--Sowmini
Tried one for tcp connections:
bash-3.00# echo "::walk ipcl_tcpconn_cache |::print -t conn_t conn_sqp | ::grep . | ::print squeue_t sq_count " | mdb -k
sq_count = 0x1071
sq_count = 0x6ce
sq_count = 0xa3d
sq_count = 0xb12
sq_count = 0x1071
sq_count = 0x53e
sq_count = 0x8bc
sq_count = 0xda3
sq_count = 0x8d9
sq_count = 0xda3
sq_count = 0x541
-ey
--
Erhyuan (EY) Tsai
Accessline: X40209/650-472-1468
Performance Engineer
PAE/Performance Availabilty, and Architecture Engineering
|
_______________________________________________
networking-discuss mailing list
[email protected]
