Re: Log pinned, filled, almost crashed
2005-11-08 14:18:31
On Nov 8, 2005, at 1:56 PM, Joni Moyer wrote:
This is a good idea, but what would reasonable settings be for the
throughputdatathreshold & throughputtimethreshold? They both have a
maximum of 99999999, but I can't imagine that you would want to
wait that
long... Any suggestions/experiences with these settings would be
appreciated. Thanks!
Settings choices for various tuning and control numbers may often
depend upon site realities, policies, and experiences, rather than
merit concrete values. The values we choose for other controls such
as Idletimeout, for example, may be more appropriately driven by
client characteristics.
I would tend to pursue implementation of Throughput* controls, if
needed, by examining the throughput and run times of prevailing
client sessions, as perhaps harvested from server-logged ANE
messages. Thereby you will derive "normal" numbers for your
environment, and be able to form numerical values to safeguard
against detrimental client behavior which derive from unreasonable
conditions.
Richard Sims
|
<Prev in Thread] |
Current Thread |
[Next in Thread>
|
- Log pinned, filled, almost crashed, Roger Deschner
- Re: Log pinned, filled, almost crashed, Richard Sims
- Re: Log pinned, filled, almost crashed, Joni Moyer
- Re: Log pinned, filled, almost crashed,
Richard Sims <=
- Re: Log pinned, filled, almost crashed, Gerald Michalak
- NDMP TOCs, John Bremer
- Re: Log pinned, filled, almost crashed, Stapleton, Mark
- Re: Log pinned, filled, almost crashed, Vats.Ashok
- Re: Log pinned, filled, almost crashed, Vats.Ashok
|
Previous by Date: |
Re: Log pinned, filled, almost crashed, Gerald Michalak |
Next by Date: |
Re: Log pinned, filled, almost crashed, Vats.Ashok |
Previous by Thread: |
Re: Log pinned, filled, almost crashed, Joni Moyer |
Next by Thread: |
Re: Log pinned, filled, almost crashed, Gerald Michalak |
Indexes: |
[Date]
[Thread]
[Top]
[All Lists] |
|
|