feigin@batcomputer.tn.cornell.edu (Adam Feigin) (06/22/88)
In article <5400026@iuvax> jec@iuvax.cs.indiana.edu writes: > > Does anyone have any experience with the new TCP3.1 just released >from Apollo. I haven't installed it yet since what we have seems to >mostly work. I'll probably install it in a week or so unless someone >tells me that it doesn't work. Well, we were running TCP3.1BETA at the Language Analysis Project (my former workplace) for several months prior to the release. IT is much better than 3.0, and seems to work without any problems. The only problem (in the beta stuff) was that if you were running X, and aborted out of the server, it would sometimes take out the routing information. The tcp_server, inetd, and routed would still be running, and you could r(sh, login etc.) to other nodes on the Domain ring, but you couldn't get out through the gateway. Pinging the gateway (ethernet side) from the Ring and the outside world would show it as dead. This is probably fixed in the release version, although I cannot vouch for that since we had not received it at LAP by the time I left 3 weeks ago. Adam
dennis@PEANUTS.NOSC.MIL (Dennis Cottel) (06/22/88)
> Does anyone have any experience with the new TCP3.1...
We have been using a beta release version of TCP3.1 (BL04) since
January, and it has been solid. We mostly use the Berkeley "r"
commands through an Ethernet gateway to a bunch of 4.3 BSD VAXes.
We are not currently using the subnetting features, although we intend
to switch in the near future. We are not running the routing daemon;
the host tables are all set in the local.txt file.
Dennis Cottel Naval Ocean Systems Center, San Diego, CA 92152
(619) 553-1645 dennis@nosc.MIL sdcsvax!noscvax!denniscoop@METO.UMD.EDU (James N. Cooper) (06/24/88)
> Does anyone have any experience with the new TCP3.1...
We have been using beta-test versions of TCP3.1 since last
October and generally each new "BL" has been better. We
do subnets and run routed. The latest version "BL08" is
very robust and does not drop remote telnet sessions like
the old version.
Jim Cooper
Dept of Meteorology
Univ of Maryland
coop@meto.umd.edubhv@igloo.UUCP (Bronis Vidugiris) (06/27/88)
We've had 1 crash of the tcpserver since we installed 3.1. We are hoping that the situation will get better when we install 3.1 on all the nodes, rather than just two of the nodes. There is still at least one problem with the server (that I intend to document and subit to Apollo Real Soon NOw) involving the receiving of a RST during one of the closing_wait states (I forget which one). Anyway, it tends to cause the tcp_server to dump core all over the place with a "heap corruption" error message. Usually, I have to reboot (with associated salvol) to get things straightend out again after this happens.
rees@A.CC.UMICH.EDU (Jim Rees) (06/28/88)
We have been using a beta release version of TCP3.1 (BL04) since
January, and it has been solid. We mostly use the Berkeley "r"
commands through an Ethernet gateway to a bunch of 4.3 BSD VAXes.
You are almost certainly violating your beta agreement by saying you
are running beta software. This isn't so bad, but someone else recently
posted a list of bugs in some beta software. This is a good way to get
yourself cut out of all future beta tests, and generally make a bad name
for yourself.
-------