Upgrade to v 2.3.2 fails with etcd error - has this been resolved?


(Ryamgithub) #1

I saw a post that to get around the etcd issue, to delete the directory - but in that post there was a reference to someone loosing their nginx config -
Has this been resolved or is there a workaround so I don’t loose my nginx config?


(Jochen) #2

Please provide some context and elaborate on your problem.


(Ryamgithub) #3

I did an upgrade this morning to my system and had the etcd error after the upgrade.
I looked through posts on this site and saw the post “Upgrade of Graylog appliance (omnibus) failed etcd issue”.

"Hi,
technically the upgrade of etcd should work out of the box. The errors in your logs could indicate a corrupted etcd database long before the upgrade. Maybe the disk ran out of space or something different.
If this is a single server setup you can delete the etcd data directory and start it from scratch:

sudo graylog-ctl stop etcd
cd /var/opt/graylog/data/etcd/
sudo rm -r *
sudo graylog-ctl start etcd
sudo graylog-ctl reconfigure
If you are running a cluster you have to repeate the set-cluster-master and refonfigure-as- steps.
Hope that helps!

Marius."

This post mentioned to delete the data in the etcd directory and that this would resolve that issue.
Later in the post someone mentioned that after doing this they lost their nginx config.
"I did the this update as well and it was a bit of a train wreck.
Deleting the ETCD files fixed my ETCD service down issue.
It also stopped a few thousand log entries a minute being logged.
Now I need to get NGINX set back up as all of it’s configuration appears to have disappeared.
At least I can log in via port 9000, but no SSL yet.
This is more of a bump then anything else so the developers hopefully see this post.

Thanks,

Keith"

Before I do the upgrade again I just wanted to check if I will loose any of my configs by deleting the files in the directory listed


(Jochen) #4

Create a snapshot before running the upgrade (as you should always do) and try it. :wink:

The nginx configuration in the OVA (or any other omnibus-based distribution of Graylog) is auto-generated anyway and you can regenerate it any time.


(system) #5

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.