Hyper-V lost synthetic NIC after Debian 8 dist-upgrade?

Question

I'm running a Debian 8 vm on Hyper-V (Win Srv 2012R2) and when I did a dist-upgrade of stable, I lost all network connectivity. Using the console, I was able to see the network interfaces as UP but could not ping any external address.

The dist-upgrade upgraded the kernel to linux-image-3.16.0-4-686-pae (3.16.7-ckt11-1+deb8u2).

I was able to work around this problem by adding 2 legacy network interfaces that are recognized with this new kernel, but I get a much reduced performance.

The Hyper-V modules seem to be loaded fine:

root@rproxy3:~# lsmod | fgrep hv_
hv_utils               17454  0
hv_netvsc              30069  0
hv_storvsc             17048  2
hv_vmbus               27978  6       hyperv_keyboard,hv_netvsc,hid_hyperv,hv_utils,hyperv_fb,hv_storvsc
scsi_mod              164132  5 sg,libata,sd_mod,sr_mod,hv_storvsc
root@rproxy3:~# dmesg | fgrep hv_
[    1.303529] hv_vmbus: Hyper-V Host Build:9200-6.2-16-0.16729; Vmbus version:2.4
[    1.336977] hv_vmbus: registering driver hv_storvsc
[    1.343214] hv_vmbus: registering driver hid_hyperv
[    1.344921] hv_vmbus: registering driver hv_netvsc
[    1.345790] hv_netvsc: hv_netvsc channel opened successfully
[    1.359193] hv_netvsc: hv_netvsc channel opened successfully
[    2.644921] hv_netvsc vmbus_0_13: Send section size: 6144, Section count:170
[    2.645032] hv_netvsc vmbus_0_13: Device MAC 00:15:5d:01:56:1c link state down
[    3.060851] hv_netvsc vmbus_0_11: Send section size: 6144, Section count:170
[    3.061034] hv_netvsc vmbus_0_11: Device MAC 00:15:5d:01:56:1b link state down
[    7.672833] hv_vmbus: registering driver hyperv_fb
[    7.893312] hv_vmbus: registering driver hyperv_keyboard
[    8.071484] hv_utils: Registering HyperV Utility Driver
[    8.071487] hv_vmbus: registering driver hv_util

but only the 2 legacy NICs are detected (I have 2 legacy and 2 synthetic NICs configured):

root@rproxy3:~# dmesg | fgrep eth
[    1.351882] net eth0: Digital DS21140 Tulip rev 32 at Port 0xe880, 00:15:5d:01:56:1e, IRQ 9
[    1.503780] net eth1: Digital DS21140 Tulip rev 32 at Port 0xec00, 00:15:5d:01:56:1f, IRQ 9
[   11.628988] net eth0: Using EEPROM-set media 100baseTx-FDX
[   11.733151] net eth1: Using EEPROM-set media 100baseTx-FDX

Any idea what to do to debug this further?

Answer 1

As a very late answer to this problem, I was never able to make it work with Generation 1 machines, but after reinstalling the vm as Generation 2, it's been working fine for several months now.