I am using Debian Jesse. My repository is set up as described in the Graylog documentation. A few weeks ago I upgraded my test instance of Graylog from v2.2.2+691b4b7 to v2.2.3+7adc951. I also updated my OS and elasticsearch. Everything went very smoothly. I just did a:
Today I tried to update my production. Everything was the same including my graylog repository in my production. After the update, I noticed Graylog did not update, but everything else updated. I downloaded the latest respository:
I noticed the directions in the documentation were for 2.2, but I just changed everything to 2.3. I then did another aptitute update/safe-upgrade. Everything upgraded with no error messages, and the services appeared to start. Nothing was listening (when viewed with netstat) however, and the error logs were not very helpful other than a message about a corrupted index rebuilding. I could not get the graylog web interface to respond. Elasticsearch was showing everything was healthy, however. I finally gave up and rolled back to a snapshot, but lost some messages. Does anyone know why an upgrade two weeks ago would be successful and today it would fail for virtually identical systems? I would like to try again, but I don’t want another failed upgrade.