Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 18 Oct 2008 01:02:34 +0200 (CEST)
From:      Wojciech Puchar <wojtek@wojtek.tensor.gdynia.pl>
To:        Aniruddha <mailing_list@orange.nl>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: ("SOLVED") Under heavy load internet gets killed, only a reboot can bring it back up
Message-ID:  <20081018005328.C13134@wojtek.tensor.gdynia.pl>
In-Reply-To: <1224249917.4053.6.camel@debian>
References:  <1224054780.4011.20.camel@debian> <20081015072630.GA70901@icarus.home.lan> <1224069478.4247.7.camel@debian> <20081015113101.GA76278@icarus.home.lan> <20081015120911.GI14769@cdnetworks.co.kr> <1224080661.5407.2.camel@debian> <20081015235638.GA22992@cdnetworks.co.kr> <1224249708.4053.0.camel@debian> <1224249917.4053.6.camel@debian>

next in thread | previous in thread | raw e-mail | index | archive | help
> I 'solved' this by buying a new nic. It's an Eminent with a RTL8139D
> chipset.

you SOLVED it not 'solved'. onboard NICs are usually crap, except high end 
motherboard.

they solder in what's cheapest in the moment. if chips are buggy, they add 
workaround in windoze drivers.

and finally - for average windoze user there is nothing strange that 
he/she needs to reboot the system many times a day. they don't even know 
why network stopped working - is it hardware problem, driver problem, 
system crash. they just reboot.


on 3 motherboards i've got with gigabit NIC, 1 are completely unusable 
(nvidia), one are barely usable after turning off txcsum and rxcsum:

re0: <RealTek 8168/8168B/8168C/8168CP/8111B/8111C/8111CP PCIe Gigabit 
Ethernet> port 0xd000-0xd0ff mem 0xf9000000-0xf9000fff irq 17 at device 
0.0 on pci4
re0: turning off MSI enable bit.
re0: Chip rev. 0x38000000
re0: MAC rev. 0x00000000



the fully working one is:

bge0: <Broadcom NetXtreme Gigabit Ethernet Controller, ASIC rev. 0x4101> 
mem 0xc0100000-0xc010ffff irq 16 at device 0.0 on pci3
miibus0: <MII bus> on bge0
bge0: Ethernet address: 00:14:5e:94:49:19
bge0: [ITHREAD]



but the last one is on IBM server.



i've just seen

re0: turning off MSI enable bit.


looks like fix recently added as i upgraded this machine to 7.1 from 6.3
already funny thing - feature that has to be disabled ;)





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20081018005328.C13134>