Big issues trying to upgrade from OD 0.7 to OD 1.1.0

General Feedback
1

Hi there,

I did an application with open distro 0.7. We are trying to upgrade (rpm version on cent OS) to open-distro 1.1.0 and we are facing big issues. Some of them are already known like the dependencies one but this is not everuthing.

I wonder, do you think it is possible to do the upgrade this way :

  • Installing a fresh open distro 1.1.0 on another machine.
  • Migrate the security plugin config files and loading them on the new 1.1.0 stack.
  • Then, simply do a reindex-upgrade-remote for all the other indices. (kibana, alerting, datas used in visualisation etc.)

Can this work ?

many thanks for your attention
Olivier

2

Hi,

the security config you can backup and restore with the following commands:

#create backup:
/usr/share/elasticsearch/plugins/opendistro_security/tools/securityadmin.sh -backup ~/opendistro_backups -icl -nhnv -cacert /etc/elasticsearch/root-ca.pem -cert /etc/elasticsearch/kirk.pem -key /etc/elasticsearch/kirk-key.pem

#restore backup
/usr/share/elasticsearch/plugins/opendistro_security/tools/securityadmin.sh -h localhost -p 9300 -cd ~/opendistro_backups -cacert /etc/elasticsearch/root-ca.pem -cert /etc/elasticsearch/kirk.pem -key /etc/elasticsearch/kirk-key.pem

The other parts should be answered by someone else who is more familiar with the elasticsearch base (index, …).

Regards

3

Thank you thsul for your answer !

I am aware of this procedure. It’s really what turns around the other indices that worries me. :wink:

best regards,
Olivier

4

big YES!

I just did this. And it was much easier than trying to backup and migrate; at least for me. That was a huge mess. Between the index multi-type issues and internal roles built in, and a few other subtle changes, it was just ugly.

  1. Started with ODFE 0.9 (ES6.7.1)
  2. initiated a new 3 node cluster ODFE 1.0.2 (ES7.1.1) and one kibana node.
    some settings were imported some were not. it was easier to just rebuild the security settings
    removed demo certs and installed prod certs, so that both clusters using same CA.
  3. Re-indexed data to new cluster - this solved the multiple-type issue for many indices
  4. Imported visualizations and dashboards from 0.9, a few adjustments.
  5. Once all data was migrated, adjusted DNS to use new kibana instance, shut down ODFE 0.9 cluster, removed ODFE 0.9, re-installed fresh 1.0.2 and added those nodes to the 1.0.2 cluster

rlk

5

Thank you for sharing your experience rlk !