wb@gamma.UUCP (06/30/87)
SR9.5.1 was, according to some at Apollo, the answer to all of our ills; the latest and greatest. Well, it was late. I remember being told at a meeting held over a year ago in Chelmsford that SR9.5 would soon be here. Is it great? I don't think so. Not only haven't they fixed any of the TCP/IP routing problems which I reported to them about a year ago, I do not see any of the performance improvements which were supposed to be associated with the recompilation of the operating system with the new compiler(s). I'm interested in hearing a discussion among other users as to their level of satisfaction with SR9.5.1. I'm now told that all of my TCP/IP problems will be fixed in SR9.6 (which will include TCP/IP 3.1), but I can't seem to find out when that will be released (next year?). It may be sooner, though, since one SR9.5.1 UCR came back with the comment that the problem in question would be fixed in SR10! Am I the olny one who is experiencing severe problems using Apollo TCP/IP is a very large networking environment? Bill Beblo Bell Communications Research 290 West Mt. Pleasant Ave., Rm 1D-148 Livingston, New Jersey 07039 (201) 740-4421
lid@cernvax.UUCP (lid) (07/02/87)
In article <401@gamma.UUCP> wb@gamma.UUCP writes: >SR9.5.1 was, according to some at Apollo, the answer to all >of our ills; the latest and greatest. Well, it was late. >I remember being told at a meeting held over a year ago in >Chelmsford that SR9.5 would soon be here. Is it great? >I don't think so. Not only haven't they fixed any of the >TCP/IP routing problems which I reported to them about a >year ago, I do not see any of the performance improvements > ... We have received 9.5.1 some times ago but not yet installed it because of problems due to recompilation of all programs etc. I was hoping that TCP/IP problems were gone in 9.5.1 but you say they are still there. We have 34 nodes running on 2 bridged rings (G700 bridge, kind of T1) and the TCP gateways are on only one of the 2 rings. The nodes on the ring without gateways still have problems in reaching anything on the Ethernet, what's funny is that sometime they work and sometime they don't, without any apparent reason. We run TCP_server on all nodes as described in the TCP manual with all the RIP_server etc., the tcpstat -g command just says what it is supposed to say etc. but ftp something_on_ether sometimes gets connected and other times says "destination not responding". All tcp_server around seem to be fine. I have given up in trying to solve the problem and was waiting for 9.5, but may be better to wait for 9.6 (I was told Sept/Oct). Has anyone this kind of problems with bridged rings ? or similar problems that could be related to this one ? Any solution, workaround ? By the way, is someone out there running multiple gateways with some sort of fallback, when the first gateway (in the host table) goes down the 2nd gets used and so on ? If yes, how did you do that ? Achille Petrilli
SofPasuk@imagen.UUCP (Munach Rvi'i) (07/03/87)
In article <401@gamma.UUCP>, wb@gamma.UUCP writes: > SR9.5.1 was, according to some at Apollo, the answer to all > of our ills; the latest and greatest. Well, it was late. > I remember being told at a meeting held over a year ago in > Chelmsford that SR9.5 would soon be here. Is it great? > I don't think so. Not only haven't they fixed any of the > TCP/IP routing problems which I reported to them about a > year ago, I do not see any of the performance improvements > which were supposed to be associated with the recompilation > of the operating system with the new compiler(s). > > I'm interested in hearing a discussion among other users > as to their level of satisfaction with SR9.5.1. I'm now told > that all of my TCP/IP problems will be fixed in SR9.6 (which > will include TCP/IP 3.1), but I can't seem to find out when > that will be released (next year?). It may be sooner, though, > since one SR9.5.1 UCR came back with the comment that the > problem in question would be fixed in SR10! > > Am I the olny one who is experiencing severe problems using > Apollo TCP/IP is a very large networking environment? No, you are not. At IMAGEN we have 50+ diskless nodes (DN300 and DN330) and 10 file servers (DSP-80a and DSP-90). We use ethernet tcp/ip to communicate with printers, our VAX systems, Sun workstations, etc. The tcp/ip that was shipped with 9.5.1 has more errors in it than an early Mets game, more bugs than an entomology lab. If suffers from data corruption and zero window problems and miscellaneous crashes. We have Apollo personnel onsite now analyzing our network - the kids in Chelmsford are allegedly waiting for the results such that they can really fix tcp/ip (3.1) and then release 9.6. Do we think that this will solve all our ills? Don't bet on it!
guzzi@uicsrd.UUCP (07/06/87)
We have had problems with the TCP server and the "spiffy" new compilers of 9.5.1 (and 9.5 - we were dumb enough to agree to be a beta-test site for 9.5). The 9.5 tcp/rcp had a tendency to scramble large files -- a bug that Apollo said they could not reproduce. The update to 9.5.1 has reduced the problem, but we still get some errors in large file rcp's. The rwhod deamon has not worked correctly since 9.2.* and our routed tables get munged occasionally and I have to kill the tcp server and related processes are restart. We too have been assured that 9.6 will fix these problems and Apollo asked if we would like to beta-test 9.6 for them. I was proud of myself for politely declining. This was only about 3 weeks ago, so 9.6 probably won't be out for a while -- if they actually test their software before they ship it. What fun! Just think, we could have had suns instead. Just another "satisfiled" Apollo user, -- Mark Guzzi U of Ill. Center for Supercomputing R & D guzzi%uicsrd@a.cs.uiuc.edu