Some days ago I mentioned a couple of problems with the eepro100-driver,
after a lot of input and testing, it seems like it's a fault in the nfs-code
somewhere.
Testing was done with one nfs-server, running linux 2.4.13 and one
nfs-client (either a Dell PowerEdge 2550 with 1GB RAM and Intel EEpro100
network card, or a Sun Enterprise 450 with 4GB RAM and hme-interface).
Running bonnie++ on the solaris-server over a nfs-mounted share:
Sun Solaris 8 [nfs-client] <-> Linux [nfs-daemon]
3Com-NIC in server: NFS-timeout after some minutes when doing "Writing
intelligently"-test. Max. throughput of 8.5MB/s
EEpro100 in server: NFS-timeout after some minutes when doing "Writing
intelligently"-test. Max. throughput of 8.5MB/s
Linux [nfs-client] <-> Linux [nfs-daemon]
3Com-NIC in server: Everything was fine, no errors whatsoever. Max.
throughput of 11MB/s.
EEpro100 in server: Everything was fine, no errors whatsoever. Max.
throughput of 11MB/s
I'd be glad to help figure out _why_ we get nfs-timeout when we do sun
solaris <-> linux. Just tell me what to run or do :)
If you need any more info about setup and tests, please tell me, and I'll be
glad to provide the details.
--
Thomas