From: Gerard Henry (ghenry@cmi.univ-mrs.fr)
Date: Wed Aug 25 2004 - 09:00:25 EDT
Many thanks for the replies. I hope I am not too late in summarizing.
i've just finished a big maintenance
my first mail wasn't so clear, because some people suggested something i have already done
- i say "Cisco switches was configured without autonegotiation 100Mb/s", but it's also right for my all suns machine according infodoc 16144 since 4 years
so it's not the way
but T. Dick send me info about a sun blue print, so i did new tests with autonegociation
my basic test consist of:
/bin/time -p dd if=/dev/zero of=/export/nfs/fich1Go bs=1024k count=1000
output file may be on local disk, or on a partition nfs mounted
if nfs server (e250 with d1000 raid 1+0) is configured with autonegociation, i saw:
Aug 24 15:32:23 kodaly nfs: [ID 333984 kern.notice] NFS server 193.50.130.112 not responding still trying
and file size stopped at 32768 bytes
and when i try to stop process, i see many errors:
Aug 24 15:39:56 kodaly nfs: [ID 664466 kern.notice] NFS commit failed for server 193.50.130.112: error 16 (RPC: Failed (unspecified error))
but if i did a ftp connction, it works!
Some results:
- time to write on a local disk (d1000): 31.6 seconds (32Mb/s)
- from a sun nfs client on this disk: 105 sec (10Mb/s)
- from a linux nfs client: 185 sec (5.6Mb/s)
all machines are 100Mb/s full duplex
from ftp, i obtain 5MB/s (40Mb/s), it's better than to write on a disk!?
and finally, ufsdump between e250 and ultra10 with LTO2 gave 5.2MB/s
very strange indeed, i don' understand what's happen, or perhaps my tests are irrelevant
gerard henry
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:29:21 EDT