Are you perhaps running on a BSD386 system? I have this same problem in my mud but ONLY after moving to the BSD386 system. Its an annoyance really. I tend to ignore it and call it an OS problem. Though it could also be something that wasnt' necessary before but probably should have been done anyway kinda thing... System call funcitonality has a tendency to change as *nix matures. At any rate, if you find the answer. Feel free to share it! #:-) BTW, a hack fix for the problem is to retry the bind but also supply a default timeout at which time the mud quits, the reboot script takes over again, and eventually, the port will clear. <cross your fingers> Also, notice on netstat and you'll possibly see a socket to your now dead mud in state CLOSING. Thats what I see when this happens to me. From what I understand <this might be wrong>, when an application that has bound a port quits, for any reason. The OS is responsible for contacting all connected sockets to that port and notifying them to close connection. Then waiting for aknowledgment.. and Closing by default after a certain timeout period. That last part, the default timeout period, seems to be missing/forgotten/failing in BSD386 systems... in my opinion. --Paul On Wed, 12 Jul 1995, Steven R. Nevid wrote: > I am having trouble with my v 2.20 mud.. > One all around problem is that when the mud reboots, it can't come right > back up, as some of the syslogs give the error "Bind already in use". It > evetually does come back up.. That is until as of late. Now when and if > it crashes the mud never comes back up always getting the "Bind already > in use" error in all the syslogs.. It stays in this "loop" until I kill > the mud in my processes. This is hopefully something maybe at least one > of you had experience with. Please repond ASAP. Thanks > > -=Steve=- > > mud address: cerberus.sunyit.edu 4000 Imps: Chaos (me) and Vall
This archive was generated by hypermail 2b30 : 12/07/00 PST