Lucene Search index rebuild

frederic.deniger · February 19, 2020, 4:02pm

Thanks. Sorry for this question but it means we have to build our own openmrs-core jar ?

ibacher · February 19, 2020, 5:06pm

Or you could contribute a patch which makes it possible to select between them .

mksd · February 27, 2020, 2:53pm

@frederic.deniger the change is to happen here. Thanks @dkayiwa for the pointers!

I don’t fully understand what’s at stake here, @ibacher, shouldn’t this always be async?

mogoodrich · February 27, 2020, 3:26pm

@mksd the downside is that while the index is rebuilding you make search for patients and not find them even though they are in the database… though there’s certainly a trade off between that and having the system take an hour to start up.

mksd · February 27, 2020, 3:29pm

Sure, that makes perfect sense.

What do you guys think about the configuration mechanism here to drive sync vs async? Because the setting has to exist before a first startup, I would lean towards a runtime property, but what do others think?

ibacher · February 27, 2020, 4:08pm

A runtime property makes perfect sense to me. I’d even think it might make sense to default the property to using async.

We might even add a MassIndexerProgressMonitor implementation so we can notify the user when the initial indexing is done, if it’s done asynchronously. Or even use a standard DB query while the index is in the process of being built… I realize that might given inconsistent results, but it could be better than finding nothing.