On Nov 10, 12:30pm, Marc Nicholas wrote:
One of my clients is complaining that performance between Solaris 2.5.1 systems (mainly Ultra 1s and 3000s) and an F210 is pretty damn awful. (The 210 is to be used for syslog, so the amount of disk writes isn't *that* high).
I got him to switch to NFS v2 and UDP to get around timeout issues, but performance still sucks.
The entire LAN fabric is full duplex 100baseT switched through Cisco Catalyst 5000s.
Can I ask a dumb question? Is his 2.5.1 box running full duplex 100BT? We experienced the same symptom with an Ultra 2. We also noticed lots of late collisions on the Cisco EtherSwitch 1200 uplink port serving the Ultra. When we switched the port to half duplex, the problem went away.
On a related topic, I spent a fruitless half hour trying to figure out how to configure our Ultra to do full duplex -- anyone know how to do it?
-- Marc Rouleau
VP and Chief Technology Officer Voice: (812) 479-1700 Fax: (812) 479-3439 Network WCS http://www.networkwcs.com
From /etc/system:
* Set Ethernet interface to full duplex set hme:hme_adv_autoneg_cap = 0 set hme:hme_adv_100fdx_cap = 1
-- Dave Pascoe | mailto:dave@mathworks.com | Voice: 508.647.7362 KM3T | http://www.mathworks.com | FAX: 508.647.7002 PGP fingerprint: 53 AD 71 88 2F AA 45 AC D0 2E 68 91 71 77 39 AF
On Mon, 10 Nov 1997, Dave Pascoe wrote:
->From /etc/system: - -* Set Ethernet interface to full duplex -set hme:hme_adv_autoneg_cap = 0 -set hme:hme_adv_100fdx_cap = 1
Much cleaner. For one of my machines, however, it still insisted on doing hdx until i told it that it wasn't capable of it.
I've not done it on many machines yet, but soon will have to implement this over a large number of machines. Thanks for the hint at the cleaner way. :)
------------------------------------------------------------- Dave Cole (DC1110) | dacole@netcom.ca Systems Administrator | dacole@vex.net | office/~dacole/ Netcom Canada | www.vex.net/~dacole/ 905 King Street West, Toronto, M6K 3G9 | phone - 416.341.5801 Toronto, Ontario, Canada, Earth, Sol | fax - 416.341.5725
On Mon, 10 Nov 1997, Marc Rouleau wrote:
-On Nov 10, 12:30pm, Marc Nicholas wrote: -> The entire LAN fabric is full duplex 100baseT switched through Cisco -> Catalyst 5000s. - -Can I ask a dumb question? Is his 2.5.1 box running full duplex 100BT? -We experienced the same symptom with an Ultra 2. We also noticed lots -of late collisions on the Cisco EtherSwitch 1200 uplink port serving the -Ultra. When we switched the port to half duplex, the problem went away. - -On a related topic, I spent a fruitless half hour trying to figure out -how to configure our Ultra to do full duplex -- anyone know how to do it?
Get ready to cry.
The only way I've been able to 'force' a Solaris/Sun box to do 100fdx against their auto-configured will is:
ndd -set /dev/hme adv_100fdx_cap 1 ndd -set /dev/hme adv_100hdx_cap 0 ndd -set /dev/hme adv_10fdx_cap 0 ndd -set /dev/hme adv_10hdx_cap 0 ndd -set /dev/hme adv_autoneg_cap 0 (once you hit this, it will immediately fall over to 100fdx)
You might consider checking first what your OS/drivers/interface thinks it can do first, with:
ndd /dev/hme 100fdx_cap (0 being not capable)
ndd /dev/hme 100hdx_cap (0 being not capable (not very likely))
the way to check if your sun box is actually doing fdx is: (and speed, and link integrity)
ndd /dev/hme link_mode (0 being half, 1 being full)
ndd /dev/hme link_speed (0 being 10, 1 being 100)
ndd /dev/hme link_status (0 being offline, 1 being link integrity alive and well)
There are a few other neat things you might like to peak at. check out:
ndd /dev/hme ?
------------------------------------------------------------- Dave Cole (DC1110) | dacole@netcom.ca Systems Administrator | dacole@vex.net | office/~dacole/ Netcom Canada | www.vex.net/~dacole/ 905 King Street West, Toronto, M6K 3G9 | phone - 416.341.5801 Toronto, Ontario, Canada, Earth, Sol | fax - 416.341.5725
-
Dave Cole -
Dave Pascoe -
mer＠world.evansville.net