I have an application that runs on a Linux box - Fedora 6. It
periodically produces information. Also have 3 other computers
running Fedora 6 that periodically check for updated information on
the first computer. One of the three is frequently getting timeout
messages - connection reset when polling for new data.
This is a new problem. THe same setup has worked flawlessly before.
I odn't claim to know what hte cause is, but I have some suspcions
that the issue is network traffic and/or firewall delays which occur
on the network that this machine number 3 is on.
I see there are lots of parameters that can be adjusted with sysctl.
My question is which of these might give number 3 more time to
complete its tasks?
FYI what the 3 client computers do is to attempt an sftp transfer of
a data file. If they are successful, they again (almost immediately)
invoke sftp to rm that data file. I mention this because most of the
timeout issues occur at this point. If there is no data file, they
just sleep for 30 seconds and try to look for the data file again.
They loop like this.
|