| To: | John Hesterberg <jh@xxxxxxx> |
|---|---|
| Subject: | Re: [PATCH] 2.5.44 CSA, Job, and PAGG |
| From: | Jan-Frode Myklebust <janfrode@xxxxxxxxxxx> |
| Date: | Wed, 23 Oct 2002 23:20:37 +0200 |
| Cc: | linux-kernel@xxxxxxxxxxxxxxx, csa@xxxxxxxxxxx, pagg@xxxxxxxxxxx, Robin Holt <holt@xxxxxxx> |
| In-reply-to: | <20021023145342.A6595@sgi.com>; from jh@sgi.com on Wed, Oct 23, 2002 at 02:53:42PM -0500 |
| References: | <20021023145342.A6595@sgi.com> |
| Sender: | csa-bounce@xxxxxxxxxxx |
I see the struct acctcsa has a ac_prid.. Are you planning on adding project level accounting to linux, or is this just leftovers from IRIX? I'm not familiar with CSA, but have some experience with using the IRIX system audit trail for accounting (satd). satd isn't handling the accounting very well when we get hard hangs or power outages, and no end-of-job record is written. Then we have no account for the cpu-usage of the non-finished jobs at the time. Does CSA handle this better, or is it still depending on jobs being finished before writing any accounting records? (a way telling it to log periodic intermidiate accounting records would be nice) -jf |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | [PATCH] 2.5.44 CSA, Job, and PAGG, John Hesterberg |
|---|---|
| Previous by Thread: | [PATCH] 2.5.44 CSA, Job, and PAGG, John Hesterberg |
| Indexes: | [Date] [Thread] [Top] [All Lists] |