Network not starting after reboot

January 25, 2018 2.2k views
Networking CentOS

restarted by droplet last night using CLI. Didn't pay much attention, because i've never had too.

Today realized I hadn't received any emails, which is that server. Couldn't load zpanel via web browser. Finally got in it through console. Restarted again and saw an error on startup saying NetworkManager-wait-online had failed.

Powered off via shutdown command. Then started again. same failure.

ifconfig eth0 shows NO ip.

through console, journalctl -xe is almost useless because it cuts off text.

stopped network via init.d, flushed dev via ip command, started networking again. nothing.

EXCEPT NetworkManager-wait-online is now running, before the flush systemctl status said it wasn't... but networking still won't start.

also: my registered email address for DO is on that server. So i won't get notifications of replies until i refresh this page. I'll try to keep watching.

4 Answers

This solution worked for me as well. Updating CentOS 7 to the latest broke networking completely in my droplet, but in my case, only the gateway parameter was missing.

With the email going out telling everyone to update their installs, there could be a lot of confused newbies out there (like myself!), so here's exactly what worked for me:

  • Check /etc/sysconfig/network-scripts/ifcfg-eth0
  • Make sure everything is there (IP, netmask, gateway)
  • Restart the network service

I'm not sure it's resolved... however, i finally checked the ifcfg-eth0 script and saw it was to dhcp. the IP wasn't in there, nor gateway, mask, dns, etc. So I manually added it even though a warning at the top of the file said not to edit, it was created on boot up by cloud something.

I made it look like the format of my other droplets, saved, and restarted network.service, bam I have internet.

Is this a centos issue after an update or did DO have an issue?

Doing a little research into this issue, it seems that CentOS 7 Droplets that are upgraded to CentOS 7.4 and cloud-init 0.7.9 can experience an issue where the Gateway is not properly set by cloud-init on boot. This also seems to be solved in our newer CentOS 7.4 Droplet image.

A bug report with information related to this issue with cloud-init can be found here:

https://bugzilla.redhat.com/show_bug.cgi?id=1497358

so after putting all the IP information back in, I can't reboot, otherwise I have to do enter it all in again.

I found this fix but it looks like we have to wait until it actually gets implemented upstream... @Surety01 I think this comment may be a temporary fix.

EDIT: I just applied the cloud.cfg edit/fix to my CentOS 7.4 droplet, and came back to confirm that it works, as I applied it and rebooted to test as well - so far so good.

Have another answer? Share your knowledge.