Hello.
Yes, i did use that tool but the cluster was already displaying the same behaviour before that with exactly the same log messages.
Those log messages appeared right after i restarted the node in 7.4.2 right after I upgraded the elasticsearch package from 6.8.3 to 7.4.2. I used the tool after when I was already out of ideas and had already downgraded to 7.3 and 7.2 with no change to my situation.
I cannot restore from a snapshot. There is no disk space in that machine to fully backup the data.
If I have to restore from the backups I have, I will lose about 1 month of data.
Is there no indication of what might be wrong on the log messages ? I can see the node being elected master and it sees all the indexes because it lists them in a gigantic log message. If i enable TRACE logging in MetadataService (i think thats the name), I can see it writing the cluster state to all indices. I don't know why it fails.
Regards,
Ricardo Ferreira