Update ELS

Diaz

We updated Elasticsearch (ELS) to version 7.1.3 according to the documentation, and then we executed the command rm -rf /usr/share/kibana/optimize/bundles/* && systemctl restart kibana after the update. However, we did not follow the further recommendations because we did not change the Kibana configuration file, and we also do not have the /opt/e-doc/ directory. The documentation mentioned that there should be a replacement of "WIKI" with "e-doc," but this replacement did not occur. Subsequent attempts to initiate the update were unsuccessful. Can you provide guidance on what might be the issue?

SzymonC

Hi,

Please check top right corner button (Wiki). Just click on it and see if wiki UI opens up. It should look somehow like this:
Please check systemctl status e-doc and systemctl status wiki.
Finally check if directories have been created: ls -lah /opt/wiki and ls -lah /opt/e-doc

Diaz

SzymonC

Checked, no changes

Diaz

Checked, no changes

SzymonC

OK, but the wiki to e-doc change was introduced in version 7.2, so if you updated to 7.1.3, then this service will not change. 🙂

Diaz

Yes, indeed, we have upgraded to version 7.1.3. Now, when running the update script ./install.sh -u, the update process goes through, but the version does not actually get updated. We want to upgrade to the latest available version and are proceeding towards it step by step, incrementally updating the versions. Currently, we are performing the updates on a test ELS (Elasticsearch) environment built from two data nodes in a cluster.

SzymonC

Ok, so let me get this right:

You now have version 7.1.3, after update.
You want to update further, to 7.2.0.
You run ./install.sh -u and script goes OK, but version isn't changing.

Did I understand that correctly?
What is the version of the package from which you are running install.sh -u?

Diaz

SzymonC Hello! Yes, our Elasticsearch (ELS) has been updated to version 7.1.3. Next, we intend to continue upgrading to 7.2.0, then 7.3.0, and finally to 7.4.0 to ensure we have the latest available version. I do run the script ./install.sh -u, and it operates normally, but the version does not get updated. The package version that the script is working with is energy-logserver-client-node-7.1.3-1.el7.x86_64.rpm

SzymonC

OK, that's what I thought.
To upgrade further, you need to download next versions of ELS. So script in package 7.1.3 is used to upgrade the system to version 7.1.3 and nothing beyond that. Script from package 7.2.0 will upgrade system to version 7.2.0 and nothing more. And so on.
Drop me an email or contact our distribution in your region and we'll send you exact links for specific versions.
After upgrades you can remove directory of /root/install, it will make process of upgrading easier.

Diaz

SzymonC Hello! Thank you for your advice; it helped us with updating ELS.

Diaz

SzymonC Good evening. We have started updating the cluster in production, and after the last update to version 7.4.0u, our Kibana is not starting because it cannot connect to Elasticsearch (OpenSearch). We ran the command "netstat -tlnup" and did not see Java listening on port 9200. What could be the reason for this?

SzymonC

In such case most important log is log from database. Please check log file in /var/log/elasticsearch/<cluster_name>.log.
Port 9200 is port used for database API, so if this is not active, then most likely database/opensearch isn't running. Please check log file and check if you have more info there. 🙂
You can also check if curl -k -u logserver 127.0.0.1:9200 is returning anything. Example response would be like below:

{
  "name" : "node-1",
  "cluster_name" : "logserver",
  "cluster_uuid" : "As7tlPTWSzyyeaVXxOnFBA",
  "version" : {
    "distribution" : "opensearch",
    "number" : "2.8.0",
    "build_type" : "tar",
    "build_hash" : "db90a415ff2fd428b4f7b3f800a51dc229287cb4",
    "build_date" : "2023-07-28T09:54:26.952266Z",
    "build_snapshot" : false,
    "lucene_version" : "9.6.0",
    "minimum_wire_compatibility_version" : "7.10.0",
    "minimum_index_compatibility_version" : "7.0.0"
  },
  "tagline" : "The OpenSearch Project: https://opensearch.org/"
}

Diaz

Hello! Yesterday, we found what the problem was. In /etc/elasticsearch/jvm.options, the variable for temporary files was specified as ES_TMPDIR instead of OPENSEARCH_TMPDIR. This prevented Elasticsearch from starting, even though the Elasticsearch status showed as active. Thank you for the feedback.