paul@msdc.UUCP (Paul Manno) (07/09/86)
We have been experiencing an irritating problem with the TCP ethernet on our 4.2BSD (BRL's) VAXen, all running DEUNAs. If there is a known 4.2 solution to the following (other than use 4.3) or if there is a helpful person out at BRL reading this, please respond by mail. It happens to 750s and 785s. Periodically, one of the systems' ethernet output gets clogged and all its ethernet output ceases. The system will continue to receive packets (rwho, etc.). However, nothing goes out the interface. A "netstat -m" tells me that there are still free mbufs and memory available. The "mon" program reports 50 packets in the device output queue and no output. An "rlogin localhost" will work normally on the failing system. Any pings to the failing system from a working one will lose all packets. Usually, this occurs without any console messages although in rare cases, the message "de0: buffer unavailable" is seen. Other systems on the net continue to operate normally. My only known cure for this condition is to reboot. Please respond by mail if you have any information (indirect pointers welcomed) about this problem. Thank you. Paul Manno Medical Systems Development Corp. ...{akgua, gatech, ihnp4, mcnc}!msdc!paul