Very slow start after upgrade to OpenNMS 30.0.0

Problem:
After upgrade from version 29.0.10 to 30.0.0 OpenNMS is started within about 1,5 hour. Before upgrade OpenNMS is started within about 5 minutes. I have about 15000 nodes in database and about 60000 interfaces. I see very slow selects in database (SELECT DISTINCT ipInterface.ipAddr FROM ipInterface WHERE ipInterface.IPADDR !=
β€˜0.0.0.0’ AND (ipInterface.isManaged != β€˜D’ or ipInterface.isManaged IS NULL)).

Expected outcome:
I cleared /opt/opennms/data catalog and set postgres connections to 100. It’s nothing change.

OpenNMS version:
OpenNMS 30.0.0
PostgreSQL 12.10

1 Like

OpenNMS is only as fast as the database. If that’s slow, everything is slow.

Database is not slow. I have separate vitual machne for postgres. 64GB of RAM, 4 cores Xeon Platinum second genaration 3GHz and storage on SSD. With OpenNMS 29.0.10 system fully starts in 5 minut with the same database. Database can perform many thousand transactions with minimal load.
After OpenNMS starts, database works fine witout bottlenecks. I can see up to 10,000 transactions per second in it.

I can confirm that 30 is a lot more resource intensive at start than 29 is/was. We have some other issues with 30 as well, but something is definitely more hungry at start. Once running it seems fine but we are fighting some graph gap issues since upgrading as well, not sure if it’s related. Just getting into diagnosing these issues today.

After upgrade to version 30.0.1 opennms starts very fast.