both fast and gigabit ethernet on a Scyld cluster
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Hans Schwengeler schweng at master1.astro.unibas.chMon Jan 13 08:22:58 PST 2003
- Previous message: Lapack testing problems
- Next message: Physical questions
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hello, I have problems with a Scyld bz27-8 system running both a fast ethernet and gigabit ethernet network. I had first only fast ethernet installed and now added a gigabit network. The master has 2 3C905 cards and 1 3C996B-T card, whereas the slaves have either 1 3C905 and a 3C996B-T card or 1 built-in Intel EtherExpress Pro/100+ and a 3C996B-T card. Now the cluster boots only the dual-processor nodes (the ones with the eepro100 drivers), whereas the single processor nodes start using the fast ethernet, but the swith to the bcm5700 driver (gigabit driver) and try their arp requests on the gigabit ethernet which is not yet configured to respond. beosetup alows only to choose eth1 (fast) OR eth2 (giga), but not both of them. If I try gigabit only (eth2 on the master), then none of the slaves start. They report unknown pci devices and suggest to added the needed info to /etc/beowulf/config.boot BUT I have added it already! So how should I proceed to make all nodes available again (preferably with the gigabit ethernet interface of course). Thanks for any help. Yours, Hans Schwengeler
- Previous message: Lapack testing problems
- Next message: Physical questions
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
