|Summary:||Multicast issues with bcm and tg3 drivers|
|Product:||[Retired] Red Hat Linux||Reporter:||Need Real Name <andrew>|
|Component:||kernel||Assignee:||David Miller <davem>|
|Status:||CLOSED WONTFIX||QA Contact:||Brian Brock <bbrock>|
|Fixed In Version:||Doc Type:||Bug Fix|
|Doc Text:||Story Points:||---|
|Last Closed:||2004-03-06 07:22:35 UTC||Type:||---|
|oVirt Team:||---||RHEL 7.3 requirements from Atomic Host:|
|Cloudforms Team:||---||Target Upstream Version:|
Description Need Real Name 2002-12-05 00:56:52 UTC
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20020823 Netscape/7.0 Description of problem: We have several dell 2650s that have dual broadcom GigE nics. We are currently only running them at 100mbit. We are having lots of problems using multicast with those NICs. The problems are not real consistent, but we have seen problems with sending multicast packets and receiving them. We have tried both the bcm and the tg3 drivers. Our application talks to the other servers using multicast, and the servers are consistently not getting the messages. We had to ultimately install intel 10/100 PCI cards so our application would function properly. Using the intel nics with the e100 driver, we see no problems with multicast. We tried several different switches (cisco, foundry) with the same result. We also tried connecting 2 machines together with a crossover cable and we still saw the multicast problems with the broadcom nics.. We used several different multicast test programs: mrecv/msend available on the internet), a java multicast test program that we also found on the net, and a multicast test program using our code. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: Run a multicast test program on 2 machines with those broadcom nics. Actual Results: Sometimes we saw issues from the multicast sending machine and sometimes we saw issues on the multicast receiving machine. The sender would sometimes only send a few packets out of the total number the test program reported sending. Sometimes it would not send any. The receiver would sometimes only see a few of the packets, or none at all. Once the intel nics were installed, we did not see a single failure. Expected Results: The test programs that we tried sent 10 packets each with a unique number, and then we should see all 10 packets on the receiving side. Since we are not sending these multicast packets over the internet, it should be a fairly reliable way of talking to a group of machines without directing packets to individual machines.
Comment 1 David Miller 2004-03-06 07:22:35 UTC
No more 7.x updates, so this is unlikely to ever be fixed.