eitan@shum.Huji.AC.IL (Eitan Mizrotsky) (12/12/89)
Hi there. I'm having a severe problem here, may be some one can drop a clue: We're running Ultrix 2.2 on VAX 785, 16 MB memory. Our NFS activity is relatively intensive: About one third of the users' home-directories are nfs-mounted, and few workstations mount our file-systems. the symptoms of the problems are as follows: the console prints thousands of identical messages: "ku_fastsend: transmit failed on de0, errno 55". Once this starts, the machine's response time becomes very bad, users who try to telnet somewhere, for example, get a message saying "No more buffers". The only cure is halting the machine, and reboot it. Lately, it happens on a daily basis. Surely, we can't go on living like that. Changing the number of nfs daemons didn't help, and I'm almost sure the problem is in the mbufs allocation mechanizm. Any help will be appreciated. Thanks in advance, Eitan Mizrotsky The Hebrew University, Jerusalem ,Israel. (eitan@hujinix).