Tell me which log you want, and I’ll get it to you. I can always use nictagadm to regenerate the error if you like.
Yes, I’m certainly willing to try out a new driver to fix the error. I’m trying to set up a point-to-point connection between two computers across the 10Gbe connection, but the 1 gigabit connection is still present and is what I’m using until I can get the 10Gbe connection up. So, as long as I don’t do anything irreparable to my system, I have no problem testing stuff out to get this fixed. Thanks, John On Jul 5, 2014, at 5:42 PM, Robert Mustacchi <[email protected]> wrote: > On 7/5/14 16:02 , John Croix via smartos-discuss wrote: >> Hi Robert, >> >> I tried what you suggested, but when I rebooted, the network services went >> into maintenance mode. Here’s the /usbkey/config file that I had in place >> before attempting nictagadm. >> [root@68-05-ca-0a-2f-c5 ~]# cat /usbkey/config >> # >> # This file was auto-generated and must be source-able by bash. >> # >> >> # admin_nic is the nic admin_ip will be connected to for headnode zones. >> admin_nic=68:5:ca:a:2f:c5 >> admin_ip=dhcp >> admin_netmask= >> admin_network=... >> admin_gateway=dhcp >> >> myricom_nic=0:60:dd:45:73:22 >> myricom_ip=192.168.2.2 >> myricom_netmask=255.255.255.0 >> >> headnode_default_gateway= >> >> dns_resolvers=8.8.8.8,8.8.4.4 >> dns_domain= >> >> ntp_hosts=192.168.1.1 >> compute_node_ntp_hosts=dhcp >> >> I issued the command “nictagadm update -p mtu=9000 myricom” which put the >> line “myricom_mtu=9000” into that file. Upon reboot is when it went into >> maintenance mode for network services. I had to remove that line from the >> console and reboot, at which point it booted up normally. > > Hi John, > > Do you happen to have the output of the service log available so we can > understand why that failed? > > I took a look at the driver and from the source code, it looks like the > default is controlled based on the presence of MYRICOM_PRIV > (http://src.illumos.org/source/xref/illumos-gate/usr/src/uts/common/io/myri10ge/drv/myri10ge.c#194). > Because that's not there, we're probably setting the MTU to 1500. > > Actually, looking further at the source code, it appears that the reason > that the nictagadm commands failed is because that driver doesn't > support having Any of its properties changed at all, so it doesn't > support changing the MTU at all. In this case, I guess we can figure out > how to change the default to make it match expectations, though it's > rather unfortunate that the driver doesn't support changing its MTU at all. > > Though, from looking at the source code it doesn't appear to be too hard > to allow it to be changed before the device is started (eg. vnics are > created over it or IP interfaces). Would you be willing to help test a > driver that had this ability? > > Robert ------------------------------------------- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125&id_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
