jqj@gvax.cs.cornell.edu (J Q Johnson) (04/30/86)
We have been having a problem with our DEQNAs on all 5 of our MicroVAX II machines (running 4.3bsd with Ultrix 1.1 holdovers). Fairly frequently, every couple of hours or so, the DEQNA hangs and needs to be reset. Ultrix has a 30sec. watchdog timer which we changed to 5sec., but even a 5sec. pause in Ethernet I/O is unacceptable to us, since we use the Ethernet quite heavily. DEC has told us that this is a firmware problem in the DEQNA and unlikely to be fixed soon. However, since I hadn't seen any mention of the problem on this list, I began to wonder whether it was in fact ubiquitous. Ergo, the questions: (1) have other sites seen this problem? (2) does anyone have a fix? Additional data: it could, I suppose, be a configuration problem. The machines are all configured for general timesharing, with Ethernet used for terminal traffic. Each has an RQDX2, an SI controller and Eagle disks, 5 to 13MB memory, and nothing else. Please reply by mail if possible. I'll summarize if there are enough responses: UUCP: jqj@cornell.UUCP or jqj@bullwinkle.UUCP Internet: jqj@cornell.ARPA or jqj@gvax.CS.Cornell.EDU BITNET: jqj@CNRLCS.BITNET
dyer@spdcc.UUCP (Steve Dyer) (05/03/86)
The firmware bug in the DEQNA is "documented" in the if_qe.c driver. We've been seeing the watchdog timer on the DEQNAs at Harvard fire fairly frequently, a couple times each day, although we haven't had the chance to correlate its occurence with anything interesting. Like you, we made the timer smaller. 5 sec. unacceptable? I guess teething on the ARPA internet with its delays during peak periods has increased my patience. -- Steve Dyer dyer@harvard.HARVARD.EDU {linus,wanginst,bbncca,bbnccv,harvard,ima,ihnp4}!spdcc!dyer