Hello, I am running 3 nodes cluster, version is 2.1.2
I dont really use streams, I just have one stream configured, who catches ware rare message…
I decided for now to upgrade just on on node (non master). I stopped all processe, run the upgrade commands
sudo rpm -Uvh https://packages.graylog2.org/repo/packages/graylog-2.2-repository_latest.rpm
sudo yum clean all
$ sudo yum install graylog-server
nad started elasticsearc, mongo and graylog processes.
The graylog gets in state RUNNING, but now I am getting hundreds of messages:
2017-02-22T11:02:33.797+01:00 WARN [DefaultStreamProvider] Unable to load default stream, tried 1041 times, retrying every 500ms. Processing is blocked until this succeeds.
MAybe reason for this is, that I have yet to upgrade the other 2 nodes ?
Or do I need somewhere to define default stream ?
Thanks for any advice.
Graylog 2.2.x brings many changes to the index management and it
does not support a rolling upgrade.
Shutdown all of your Graylog nodes, upgrade the master node and start it, and after all migrations succeeded, you can upgrade and start the secondary Graylog nodes.
Also make sure to read
Ok, Thanks for the hint. But we would not like to loose any messages, while cluster is down. We will do it during night to minimaze the loss of data. Any other idea ?
Any other idea ?
This totally depends on how you’re ingesting messages and what your current infrastructure is.
I tested on test system and journal system saves messages while graylog servers are down. So I will try the same on porduction, I think by the currett load ( 50 msgs/sek) it should b eno porblem even if takes few hours to upgrade.
Just an update on this issue. I upgraded also the master node and stopped the graylog process on other 2 nodes and now the production is up and runing on latest version.