From: CRDGW2::CRDGW2::MRGATE::"SMTP::CRVAX.SRI.COM::RELAY-INFO-VAX" 7-AUG-1990 12:33:22.07 To: MRGATE::"ARISIA::EVERHART" CC: Subj: Problems With Ethernet Concentrators And Kernal Mode Software.... Received: by crdgw1.ge.com (5.57/GE 1.70) id AA16717; Tue, 7 Aug 90 09:12:39 EDT Received: From REMOTE.DCCS.UPENN.EDU by CRVAX.SRI.COM with TCP; Mon, 6 AUG 90 19:21:25 PDT Received: from XRT.UPENN.EDU by remote.dccs.upenn.edu id AA05041; Mon, 6 Aug 90 22:18:21 -0400 Message-Id: <9008070218.AA05041@remote.dccs.upenn.edu> Date: Mon, 6 Aug 90 22:21 EDT From: "Clayton, Paul D." Subject: Problems With Ethernet Concentrators And Kernal Mode Software.... To: INFO-VAX@SRI.COM X-Vms-To: @INFOVAX,CLAYTON Its been a little while so I thought the following would be of interest. 1. The recent series of messages concerning the 'support' contracts for third party software versus 'compatibility' with X version of VMS is getting much trickier. While I admit that some packages I have used, both DEC and non-DEC, make use of standard AND UNDOCUMENTED features, this is not the major problem! The major problem here results from the breakup of VMS that was done in VMS 5.0, and continues into the future. Each major code group has now been lumped together in a logical order and is loaded one after another at boot time. EACH of these code groups can CHANGE at any rev. update of VMS. These groups can be seen on a live system using the 'ANALYZE/SYSTEM SHOW EXECUTIVE' command, or the 'ANALYZE/IMAGE SYS$LOADABLE_IMAGES:*.EXE' command. ANY software that is linking with kernal privs and playing around with VMS structures is trapped with certain revs of these images. Any changes and the program needs a re-link. Unless you have complete, and accurate, sources to build from, having a support contract from the vendor is the way to get the new programs. The 'reality' of VMS today (VMS 5.x), is that nothing is 'stable' and it was designed that way. The bottom line therefore is, get and maintain a service contract to allow continued use of special software to your systems. 2. We have been fighting a major network problem for the past couple of months. It boils down to a new series of Ethernet Concentrators made by a company called Synoptics. The old boxes, Series 1000's, work fine and we have five of them. Then we bought two new boxes, 3003 and 3030 to handle expansion. These new boxes allow port level diags. as well as automatic port isolation if the port goes into blabber mode. Well the problem we have been having is with the Ethernet option of signal retiming. Not to bore you with details, but it does not work on the 3XXX's boxes. It also screws up the 1000 boxes as well. The net effect, at least in our case, is no data corruption, just traffic that 'feels' like it is going slow motion. Cluster members will drop out and back in, nodes won't reboot if they are not on the same concentrator as all the other nodes, LAT has a delay factor in the seconds. The corp. boys/girls of Synoptics are looking into the problem, but nothing in the works to solve it. The bottom line is that if you have Synoptics 3XXX boxes, you may be having the problems and should call Synoptics to add more heat. If you have other Concentrators you should ask how 're-timing' is handled if the boxes are put into a tree structure. Having fun.... pdc Paul D. Clayton Address - CLAYTON%XRT@RELAY.UPENN.EDU Disclaimer: All thoughts and statements here are my own and NOT those of my employer, and are also not based on, or contain, restricted information.