Bolt service unexpectedly stops and restore action kicks in the backend

Hi ,

I have been uploading data to Neo4J 3.5.6 community edition server running on an Ubuntu 18.04 server. At this point, I have around 1.7 million entities and ~200K relations. I also have elastic search APOC attached with NEO4J.

Recently I see that during data upload or even reading, bolt access suddenly stops even though Neo4Jservice is still running. The browser also becomes inactive. When I look at the
logs I see the following lines.

I see this recovery actions kicks in at random times and have no idea what is happening and did not see any related settings in my configuration file.

I was thinking this was happening during some data upload, but it also happened even when the upload/read from the graph was completely idle. And this happened a few times intermittently. I am uploading data through python connector (python-cypher). This is the first time this has happened and during the upload until 1.6 million entities, there were no issues at all.

Any leads to this sort of issue would be appreciated.


Can you share exactly what you're doing for import.
And also your /var/log/neo4j/debug.log

I am importing through JSON file and not doing the bulk upload with CSV. The upload is sequential with one entity at a time.

What I have pasted above is a screenshot of /var/log/neo4j/debug.log when the issue happened.

But that's from startup, not during operations ...

Can you share the whole log? And more details on what you're doing.