Interface-update event during Nethserver install over CentOS on GCP

Hi guys,

motivated by @anthony post Google Cloud Engine Compatible Image, I’ve tried to install Nethserver on Google Cloud Platform over a CentOS installation.

After launching the command


the instance become not longer accessible via ssh.

The last log reports

Jun  1 20:11:59 test esmith::event[2221]: Action: /etc/e-smith/events/runlevel-adjust/S95create-default-accounts SUCCESS [0.079828]
Jun  1 20:11:59 test esmith::event[2221]: Event: runlevel-adjust SUCCESS
Jun  1 20:11:59 test esmith::event[1787]: Action: /etc/e-smith/events/system-init/S90system-adjust SUCCESS [15.659357]
Jun  1 20:11:59 test esmith::event[2341]: Event: interface-update 

There is someone know what command performs interface-update event and if this to something to firewall rules or change ssh port?

The instance of the virtual machine after this remain unreachable also after a reset (reboot).

Thank you

had a similar problem when I did the same on my personal server (hosted on a VPS)
AFAIR, during that event, the ipv4 stops working (ipv6 keeps working)

I remember a post of mine on Nethesis italia facebook group

1 Like

Thank you @zamboni! I’ve found it.

It was date one year ago and at the end of it @giacomo talks about a bug on the way of resolution but without any other updates.

There was someone?

The old bug is already fixed.

I think this is a different problem.
Have you got console access to the machine?

Hi @giacomo

I’ve console access until in /var/log/message appear the log about the event interface-update
After this I must reinitialize a new istance of CentOS (fortunately it does not take much).

If you want, I can create a new instance of CentOS and than performe a new Nethserver Instalation over it, the best should be doing it with a more verbose log. There is an option to do this?

I’m wathing also /var/log/nethserver-install.log who is just the standard output of the command

# nethserver-install

There is another log file to watch?

I’m waiting your indication… :slight_smile:

Before nethserver-install, does the machine have any IPv4 address assigned? What does ifconfig -a says?

@davidep this is the ouput of ifconfig -a

    [stefano@nethtest ~]$ ifconfig -a
eth0      Link encap:Ethernet  HWaddr 42:01:0A:F0:CD:FD  
          inet addr:  Bcast:  Mask:
          inet6 addr: fe80::4001:aff:fef0:cdfd/64 Scope:Link
          RX packets:240 errors:0 dropped:0 overruns:0 frame:0
          TX packets:278 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:52998 (51.7 KiB)  TX bytes:30774 (30.0 KiB)
lo        Link encap:Local Loopback  
          inet addr:  Mask:
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:65536  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:0 (0.0 b)  TX bytes:0 (0.0 b)

To see how network configuration is probed and persisted to networks DB, let’s try to exclude interface-update from system-init:

rm /etc/e-smith/events/system-init/S95nethserver-base-interface-update

Then run nethserver-install. At the end, let’s see what say

db networks show
route -n
ifconfig -a 

@giacomo do you think there are other useful commands here?

Before run nethserver-install in my /etc/e-smith/ is not present system-init directory.

This is what I have…

[stefano@nethtest events]$ pwd
[stefano@nethtest events]$ ls
nethserver-base-update  nethserver-release-update

ooops :smile: sorry, that stuff is not installed yet

Edit the nethserver-install to skip the command on line 104

logexec "/sbin/e-smith/signal-event system-init"

Run nethserver-install, remove the symlink and finally run signal-event system-init.

Anyway (thanks to @giacomo) you’ll probably hit a routing problem due to mask, which is not configured correctly, thus the VPS becomes unreachable the first time interface-update is executed.

You could create /sbin/ifup-local executable script that fixes the routing table each time an interface is ifup-ed. See /etc/sysconfig/network-scripts/ifup-post for details.

Great @davidep!

Until the comment of line 104 of nethserver-install all is clear :wink:

For the next step I need some more info… :slight_smile:

Which symlink I should remove?

Did you have some resource to indicate for going in deep to this topic?

For now, however, thank you very much (also to @giacomo) :wink:

rm /etc/e-smith/events/system-init/S95nethserver-base-interface-update

The GATEWAY setting in ifcfg-eth* must point an IP in the same subnet. Now you have a single IP subnet, so it does not work. The solution would be setting up a static route, but /etc/sysconfig/network-scripts/route-* is overwritten by interface-update event.

Thus, as last resort, /sbin/ifup-local seemed viable to me…

Mission completed… soon a report of the steps needed to run a NetServer instance on Google Compute Engine.

Just one impression. Google Compute Engine management interface is more clean than the one of Amazon EC2… but NethServer installation on it is a little more tricky…

Thank you @davidep for the support…


Good job @etino I do want it, keep us updated!

1 Like

Nethserver on Google Cloud Platform… done :stuck_out_tongue_winking_eye: