Peter Dowson
01-03-2005, 07:36 AM
On 12/26/2004 2:58:29 PM, Jan Geurtsen wrote:
>Hi all,
>
>Same problem over here after
>updating fsuipc 344 and widefs
>644.
Please try manipulating some of the additional WideClient.INI parameters:
1. Increase "ApplicationDelay" from 0 to 6, or other values between
2. Increase "SendScanTime" from 10 to something bigger. You may be able to
reduce the ApplicationDelay if this is higher.
3. Increase "MaxSendQ" from 100
4. Change "WhenMaxSendQ=Recon" to "Flush"
But the last two are "avoidance" not "fixes" -- if the maximum send queue
recorded at the end of the log after close if high then something is still
wrong. 2-10 is good. 100 or more is definitely bad.
Please experiment to see if you can find something which works well.
I now have about four users with the SendQ over 100 problem and I am thinking
of changing the defaults in WideClient, again. However, I don't weant to
destroy the performance gains I've made recently, so I'm not rushing into
anything. Since I cannot reproduce the problems I am trying to gather data
from each of those who can reproduce these problems, please.
Thanks!
Regards,
Pete
>Hi all,
>
>Same problem over here after
>updating fsuipc 344 and widefs
>644.
Please try manipulating some of the additional WideClient.INI parameters:
1. Increase "ApplicationDelay" from 0 to 6, or other values between
2. Increase "SendScanTime" from 10 to something bigger. You may be able to
reduce the ApplicationDelay if this is higher.
3. Increase "MaxSendQ" from 100
4. Change "WhenMaxSendQ=Recon" to "Flush"
But the last two are "avoidance" not "fixes" -- if the maximum send queue
recorded at the end of the log after close if high then something is still
wrong. 2-10 is good. 100 or more is definitely bad.
Please experiment to see if you can find something which works well.
I now have about four users with the SendQ over 100 problem and I am thinking
of changing the defaults in WideClient, again. However, I don't weant to
destroy the performance gains I've made recently, so I'm not rushing into
anything. Since I cannot reproduce the problems I am trying to gather data
from each of those who can reproduce these problems, please.
Thanks!
Regards,
Pete