CODEWITHSUNDEEP
neo4j
user3565139
user3565139

Reputation: 1

NEO4j 3.0.4 not shutting down

I've been working on a neo4j CE db locally in version 3.0.0 creating all nodes and relationships. When time came to upload the data to heroku/graphenedb the supported version was 3.0.4 so i upgraded from 3.0.0 -> 3.0.4 locally.

After the upgrade the db started behaving strange. When shutting down using the CE osx client it hanged on shutting down. Eventually after a few hours I killed the process after checking the log, which stated that it had shut down. The logs at this point state (recreated incident to get log, hence recent date):

2016-08-31 13:05:21.396+0000 INFO [o.n.k.i.f.CommunityFacadeFactory] Shutdown started

2016-08-31 13:05:21.397+0000 INFO [o.n.k.i.f.CommunityFacadeFactory] Database is now unavailable

2016-08-31 13:05:21.424+0000 INFO [o.n.k.i.t.l.c.CheckPointerImpl] Check Pointing triggered by database shutdown [129987]: Starting check pointing...

2016-08-31 13:05:21.424+0000 INFO [o.n.k.i.t.l.c.CheckPointerImpl] Check Pointing triggered by database shutdown [129987]: Starting store flush...

2016-08-31 13:05:21.429+0000 INFO [o.n.k.i.DiagnosticsManager] --- STOPPING diagnostics START ---

2016-08-31 13:05:21.430+0000 INFO [o.n.k.i.DiagnosticsManager] --- STOPPING diagnostics END ---

Then nothing more... The only thing getting the db to run again is deleting the transaction logs. The same issue appear on the graphenedb plugin on heroku.

How do I save the data and get it into a stable state running neo4j v3?

Upvotes: 0

Views: 186

Answers (1)

Mattias Finné
Mattias Finné

Reputation: 3054

I think the best thing you could do at that point after shutdown when you observe that the process has hung for a while, is to do a thread dump. A java process will stay alive as long as there are at least one non-daemon thread alive - which could be the cause here.

Upvotes: 0

Related Questions