Arista vEOS not booting from exported configs

[Fozzie--Bear]

I'm having issues getting vEOS devices to boot from an exported configuration after wiping. I see the clock\timer appear next to the node while it's booting after a wipe, but the config itself doesn't ever appear to be applied, I just end up with a blank config.

- I've exported the configuration, and verified that I see it when looking at Startup Configs.
- Verified that nodes are set to boot from exported configuration.
- Tried both with a untouched image (going into Zerotouch Provisioning after wipe), as well as an image that has 'zerotouch disable' commited to the base image. Neither appear to work.

I see some notes in the changelog for v2.0.3-95 referencing vEOS config export...is there anything else I'm missing to get this working?

EVE-NG version: 2.0.3-95
vEOS version: 4.21.1.1F

[Uldis (UD)]

please check if logging console is enabled in running confg...
Disabling logging console makse sense boot from exported...
And better come tive chat to talk with eve experts for advice.
We can look at your config to see what is the cause of it

Uldis

[Fozzie--Bear]

please check if logging console is enabled in running confg...
Disabling logging console makse sense boot from exported...
And better come tive chat to talk with eve experts for advice.
We can look at your config to see what is the cause of it

Uldis

Was able to look at this a little more today. During the failures I was seeing the following in the logs:

Code: Select all

mount: special device /dev/nbd0p1 does not exist
umount: disk: not mounted

Looking at the veos_diskmod.sh file and running the script line by line, I encountered no errors...which lead me to think this was a timing issue. It looks like the qemu-nbd may be executing before the module is fully loaded or /dev/nbd devices are available on my system. Just dropping in a short sleep at the following location now has the wipe\restore function working 100% of the time on my system now.

Code: Select all

root@eve-ng:/opt/unetlab/scripts$ diff -u veos_diskmod.sh.bkp veos_diskmod.sh
--- veos_diskmod.sh.bkp 2019-03-28 21:23:40.591843192 -0700
+++ veos_diskmod.sh     2019-03-28 21:34:15.478404865 -0700
@@ -2,6 +2,7 @@
 cd $1
 rmmod nbd
 modprobe nbd max_part=16
+sleep 1
 for i in $(seq 0 15)
        do /opt/qemu/bin/qemu-nbd -c /dev/nbd${i} *.qcow2  && break
 done

I'll look around when I get a chance for a more elegant way of checking that the module and /dev/nbd devices are ready before attempting to attach the disk to them.

[Uldis (UD)]

can you share your config of arista to test on my eve?
you can send to:
uldis@eve-ng.net