topic bonding on DL360 RH AS2,1 in Operating System - Linux

bonding on DL360 RH AS2,1

Jeroen Peereboom — Mon, 22 Mar 2004 09:07:35 GMT

L.S.

I think I really need some good advice here:

OK, I re-installed RH AS2.1, and managed to install the Proliant Support Pack by hacking rhconfig.h: removed the SUMMIT lines, see http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=513249.

I'm trying to setup bonding.
Did the folowwing:
============
1. /etc/modules.conf:"
alias parport_lowlevel parport_pc
alias eth0 bcm5700
alias eth1 bcm5700
###options bcm5700 line_speed=100 full_duplex=1 auto_speed=0
alias bond0 bonding
options bonding miimon=100 mode=1
alias scsi_hostadapter cciss
alias scsi_hostadapter1 cciss_2445
alias usb-controller usb-ohci

============
2. /usr/sysconfig/network-scripts/ifcfg-eth*:
DEVICE=eth0
BOOTPROTO=none
ONBOOT=yes
TYPE=Ethernet
USERCTL=no
MASTER=bond0
SLAVE=yes

=========
3. ifcfg-bond0
DEVICE=bond0
BROADCAST=192.168.1.255
IPADDR=192.168.1.11
NETMASK=255.255.255.0
NETWORK=192.168.1.0
ONBOOT=yes
USERCTL=no
PEERDNS=no
BOOTPROTO=none

=================
service network stop
service network start

Simple network: 1 Linux server, 3com OfficeConnect switch, 1 laptop.

Start ping from server to PC, removed eth0, eth1: all looks well.

Start ping from PC to server, removed eth0, eth1: fails, unless there is a ping from linux server to laptop too!
Tailing /var/log/messages I see messages like:
bond0 seeing ethX is down, making ethY active.
Seems OK to me.

My questions are:
a. Is the setup correct
b. If I use service network stop / start I see complaints:
- miitool not working (but if I execute mii-tool I get decent responses)
- enslaving ethX, but 'Something broke getting the master's IP address, broadcast, ...'
What's the cause of this?

See also the attached /var/log/messages.

JP.

Re: bonding on DL360 RH AS2,1

Jeroen Peereboom — Tue, 30 Mar 2004 01:02:58 GMT

Anybody?

Re: bonding on DL360 RH AS2,1

Colin Topliss — Tue, 30 Mar 2004 06:24:53 GMT

Hi,

the error concerning mii_tool is because the bonded driver has no mii support (note from the source code):

+ /*
+ * miimon is set but a bonded network driver
+ * does not support ETHTOOL/MII and
+ * arp_interval is not set. Note: if
+ * use_carrier is enabled, we will never go
+ * here (because netif_carrier is always
+ * supported); thus, we don't need to change
+ * the messages for netif_carrier.
+ */
+ printk(KERN_ERR
+ "bond_enslave(): MII and ETHTOOL support not "
+ "available for interface %s, and "
+ "arp_interval/arp_ip_target module parameters "
+ "not specified, thus bonding will not detect "
+ "link failures! see bonding.txt for details.\n",
+ slave_dev->name);

Sort of suggests a config problem.

The reason mii_tool works on the command line is because you are doing it against the network card itself (which *does* support mii_tool).

You could try this manually by following the instructions:

"Usage: ifenslave [-adfrvVh] < [metric ] > ...\n"
" ifenslave -c master-interface slave-if\n"
"\n"
" To create a bond device, simply follow these three steps :\n"
" - ensure that the required drivers are properly loaded :\n"
" # modprobe bonding ; modprobe <3c59x|eepro100|pcnet32|tulip|...>\n"
" - assign an IP address to the bond device :\n"
" # ifconfig bond0 netmask broadcast \n"
" - attach all the interfaces you need to the bond device :\n"
" # ifenslave bond0 eth0 eth1 eth2\n"
" If bond0 didn't have a MAC address, it will take eth0's. Then, all\n"
" interfaces attached AFTER this assignment will get the same MAC addr.\n"
"\n"
" To detach a dead interface without setting the bond device down :\n"
" # ifenslave -d bond0 eth1\n"
"\n"
" To set the bond device down and automatically release all the slaves :\n"
" # ifconfig bond0 down\n"
"\n"
" To change active slave :\n"
" # ifenslave -c bond0 eth0\n"
"\n";

That will tell you if there is a general problem or just one in the config.

Col.

Re: bonding on DL360 RH AS2,1

Jeroen Peereboom — Wed, 31 Mar 2004 08:35:15 GMT

Col,

thanks for your reply!

I followed the manual steps, no messages in my terminal session; I forgot to check the /var/log/messages and I did not test by unplugging cables.
(By the way: in my original setup the tests fail because I use 1 switch only which does not known that another port than the one that just died has the same MAC address. That's why the tests succeed if I ping from the server.)

If I understand you correctly your statement is that the bcm5700 driver does not support miitool.

I cannot verify that. I hope to talk to an HP employee in due course. I let the case rest for now.

JP.

Re: bonding on DL360 RH AS2,1

Colin Topliss — Thu, 01 Apr 2004 17:29:34 GMT

Hiya,

Its the bonding driver that doesn't support mii-tool (at least according to the comments in the code). From the look of the log you attached, its failing at that point. The reason why you get a response from mii-tool when you run it is that I suspect that you are running it against the interface. Thats what leads me to think its a config problem, and why I suggested running through the steps manually. It might throw some light onto the problem.

Although I use different hardware, I'll take a look at bonding pretty soon - if I turn anything else up I'll post it here for you.

Col.

Re: bonding on DL360 RH AS2,1

Brian Landers — Mon, 03 May 2004 12:18:18 GMT

FYI, I ran into this same issue with RHEL 3.0. I was able to solve it by NOT installing the HP-supplied bonding driver. Once I went back to using the built-in kernel bonding.o driver (which required reinstalling some packages since the HP RPM doesn't uninstall gracefully!) , everything worked beautifully.