Stuart Henderson wrote: > On 2019-05-15, Predrag Punosevac <punoseva...@gmail.com> wrote: > > Hi, > > > > I am having an issue with a single 10 Gigabit interface on one of > Intel > > Xeon D-1541 network servers. Namely after the reboot the interface > > appears to be down even with a static route > > > > phobos# ifconfig ix0 > > ix0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500 > > lladdr ac:1f:6b:19:f7:72 > > index 1 priority 0 llprio 3 > > groups: egress > > media: Ethernet autoselect > > status: no carrier > > inet 128.2.204.160 netmask 0xfffffc00 broadcast 128.2.207.255 > > > > The only thing I can see is > > > > ix0: CRITICAL: EXTERNAL PHY OVER TEMP!! PHY will downshift to lower > pow > > er state! > > Looking at the driver it looks like this is a high temperature alarm > coming from the transceiver (PHY) passed on by the nic. The driver > attempts to powers down the PHY in this condition, presumably to try to > avoid damage. > > Is the cooling in this system working correctly? > > Do you still see it if you power it off for a while and let it cool > down? > > (10GBase-T is relatively power hungry.)
Hi Sten, The network interface does come after a cold reboot (complete power off not just reboot command). I replaced network cable and made sure that university network guys don't have some of DHCP server "enterprise features" on. Breaking into UEFI is not very useful but I am logged into IPMI to a two identical SuperMicro X10SDV-TLN4F servers. One of them has that problematic ix0 at pci3 dev 0 function 0 "Intel X552/X557-AT" rev 0x00: msi interface The only difference I see is that problematic server run 4 Celsius degrees warmer (75 instead of 71) but according to the limits I should be ok up until 95. Little more search reveals that I am not the only one who got hit with this thing https://tinkertry.com/how-to-work-around-intermittent-intel-x557-network-outages-on-12-core-xeon-d I will have to think through before I decide what to do. Thanks for heads up. Predrag P.S. I was ready to fire up Linux live CD in order to try to reproduce the problem and see if the Intel guys have pushed some changes into the Linux version of the driver which is not shared with this community.
