Ask Questions and Find Answers

Konrad Bijoch
We're using Liferay Portal version 7.0.4, there are two application instances working in a cluster and using the same database. There is a strange issue however. After updating some templates, only one instance responded to changes while the other didn't. We've cleared all caches and reindexed everything, but issue still persists.

With proper configuration (make extra extra extra sure that both instances use the same database) this shouldn't be the case, especially given cache-clearance and reindexing. Cross check this by shutting down the database and see if this has an effect on both instances.

Very often, behavior like this was traceable to a wrong cluster setup, despite the claim that the cluster was - of course - set up correctly. Try this by only running one server at a time. Even if both share a single search server (they do, right?), try reindexing when only one of them runs.

Konrad Bijoch
We've recently migrated the database from earlier version of Liferay and the moved the dump to another environment. Some things didn't go well however and we had to do some updates directly on the database. But those changes were detected by both instances so there was no problem here. Any clue of what is happening will be appreciated.

I might sound like a broken record: "we had to do some updates directly on the database" sounds like a good candidate. As I always say, this is a recipe for disaster. Which changes you did, we don't know (as you only mention "some").  I've seen things work apparently after such changes, but they were actually broken in a very subtle manner.

Try the options given above and report back.