Re: How do narrow down which process is causing high I/O wait times?
From: Richard Pettit [SE Toolkit Author] (richp_at_setoolkit.com)
Date: 08/28/03
- Next message: Shmuel (Seymour J.) Metz: "Re: Mad Hatter Plans OS/2 Tea Party"
- Previous message: Gary Mills: "How to get reverse DNS with a cable modem?"
- In reply to: John Davis: "Re: How do narrow down which process is causing high I/O wait times?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Thu, 28 Aug 2003 02:54:12 GMT
John Davis wrote:
> Thanks Rich for your response.
>
> I'm just curious how much does a large TCP send/receive buffer queue
> (via netstat -n) factor into a high I/O wait time?
>
> Does the writing to the TCP sockets factor into a wait time and if so
> does it have a higher priority than say writing to disk?
The cpu_sysinfo structure has a wait member that is an array with elements
W_IO
W_SWAP
W_PIO
but as I recall from my Sun days, only W_IO is ever updated which is disk I/O.
There is also a cpu_syswait structure with members
iowait
swap
physio
but the second 2 members are commented as unused, leaving disk I/O.
I have no reason to believe Solaris will report on anything but disk I/O wait.
Those with kernel source can overrule me.
Rich
- Next message: Shmuel (Seymour J.) Metz: "Re: Mad Hatter Plans OS/2 Tea Party"
- Previous message: Gary Mills: "How to get reverse DNS with a cable modem?"
- In reply to: John Davis: "Re: How do narrow down which process is causing high I/O wait times?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
