I installed the s3 snapshot plugin on my ES 5.6 cluster. Those nodes are CentOS 7.4. I installed ES using yum. I needed to upgrade to v6, so I wanted to make a backup.
I installed the repository-s3. I read the documentation for it (inadvertently, I read v6's docs).
I installed the keystore (elasticsearch-keystore create) and added the s3.client.default.access_key and s3.client.default.secret_key values to it.
I know now that on 5.x, those values should have been on the .yml config, but I made that mistake by not paying attention to the docs version.
Anyway. I left that there, didn't make the backup and tried to do a rolling upgrade to 6.x.
When I restarted the node, it failed to start giving out an error reading like: "could not create /etc/elasticsearch/elasticsearch.keystore.tmp. Check permissions". I do not have the exact text for the error, but that was the gist of it.
I had to rollback to a snapshot of that VM, to get everything to the safe side and forgot to copy the error message. I panicked. Sorry.
Is there a way I can delete the keystore from that install (after removing repository-s3, of course), so I can retry the upgrade to 6?
You can just delete the keystore file at /etc/elasticsearch/elasticsearch.keystore and recreate once upgraded. I will see if I can reproduce this upgrade issue.
I didn't exactly delete it, but I renamed it to /etc/elasticsearch/elasticsearch.keystore.BAK, after deleting the 2 key/value pairs in it, and restarted the server, to no avail. It still gave the same error. That is my recollection anyway, as I haven't tried the upgrade again.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.