hunt@ENH.NBS.GOV ("CRAIG HUNT") (01/16/89)
As I previously stated on these lists my computer is still having problems receiving routes from milnet. I recieved a few letters back stating that other people are also seeing some problems. In addition a request was made that I supply more details about my specific problem. Here are the details: The computer is a Vax 8250 running VMS 4.7 and the Wollongong 3.2 TCP/IP package. We are connected to milnet IMP 19 on port 0 using standard X.25 service over a 56 Kb line with an ACC ACP6250 interface on the Vax. We upgraded to the lastest gated from Wollongong on December 20th. The routing problem started at the end of November and was intermittent until installing the new gated on December 20th. At that time the routing problem appeared to go away but we continued to monitor the routing table every day. On Tuesday, January 10th the routing table appeared ok but when the table was checked on Wednesady the 11th there were no routes via milnet. Unfortunately we did not have tracing on for gated. The NOC and others were contacted to see if anything had changed on the network or if others were reporting problems. The night of the 11th the vax was rebooted with tracing on gated. The routes flowed in from milnet but by the next morning there were no milnet routes in the table. The trace was reviewed and showed that several routing update packets were received from 26.1.0.40. The last update was received at 22:51 on Wednesday the 11th. The protocol continued to exchange hello/I hear you messages and NR poll messages. We would receive polls from 26.1.0.40 and send replies. After three unanswered NR polls we dropped 26.1.0.40, acquired 26.1.0.65, sent 3 unanswered polls, acquired 26.3.0.75, sent 3 unanswered polls, acquired 26.1.0.40 ... You get the picture. We haven't received an update since Wednesday the 11th, despite restarts and reboots. The problem does not appear to be related to any one core gateway because they all give the same result. I suspected X.25 because we saw many errors on dda0, however on Friday the 13th (oh no!!) the NOC applied 3 X.25 patches to our PSN and the errors on dda0 dropped to nothing. I contacted Wollongong on Friday after the NOC patches had no effect on the routing problem but have been unable to connect with anyone because of the holiday weekend. I am now pursuing Mike Brescia's idea that the kernal code may need to be updated to supported the larger egp packet size in the same way that gated was updated. If anyone out there has a TWG system which is successfully receiving EGP routing updates, please let me know how you are doing it. Below are the EGP lines from my gated.conf. Thanks for any help, ---Craig ============================= # # setup EGP configuration # autonomoussystem 49 egpmaxacquire 1 egpneighbor 26.1.0.40 egpneighbor 26.1.0.65 egpneighbor 26.3.0.75 # # egpnetsreachable 129.6.0.0 26.0.0.0
craig@CWH.CAM.NBS.GOV (Craig Hunt) (01/20/89)
The milnet routing problem I reported earlier on this list may be largely resolved. (Note my caution. Overconfidence tripped me up before.) If progress has been made it has largely been the result of advice I received on this list and the help of Wollongong support staff. The version of gated which Wollongong is currently providing resolved the problem I had with receiving routing updates. The updates began to flow which moved us to a new phase of the problem. In the second phase of the problem we noted that the size of the routing table varied widely. By looking at the egp trace we determined that if we aqiured different neighbors we got very different routing updates. We were still using the old LSI-11 gateways so we "re-homed" to the BMB gateways. This was done this morning and so far, so good. We seem to have complete routing tables and to be keeping them. Wish us luck. ---Craig