[Spread-users] seeing server failures at the client
bplist at thinkpink.com
Thu Apr 21 18:39:51 EDT 2005
I'm fairly new to spread, so this may be a naive question:
We're running a fairly busy spread setup with a small number (10's) of
clients. The other day, someone ran some test code that flooded our
server with connection requests. He then rebooted the server and
restarted the daemon. No big deal.
But none of the clients (Java & Perl) got any kind of error at all,
even after the server was restarted. They just stopped seeing any
messages (over hours). My assumption was that all the clients would
get their connections reset when the server was restarted and they'd
all throw, but at least in this case that didn't happen.
To catch problems like this in the future I'd like to wrap the
SpreadConnection with another class that re-tries on catastrophic
failures. I've tried to reproduce the problem, but whatever I do seems
to result in clients getting exceptions "immediately."
Anyone ever seen anything like this?
More information about the Spread-users