Async executor and multitenancy

Flowable Engine
1

Hello,
in my spring boot app I use MultiSchemaMultiTenantProcessEngineConfiguration to implement multitenancy.
I have some rabbit listener which listens when a new tenant exists and registers it with registerTenant method of MultiSchemaMultiTenantProcessEngineConfiguration.class.
I was wondering if there we any sequences if registerTenant runs twice (e.g. by different threads) for the same tenant and datasource.
I use

config.setDatabaseSchemaUpdate(AbstractEngineConfiguration.DB_SCHEMA_UPDATE_TRUE);

so I guess the schema will be ok, but how about the async executor (I have ExecutorPerTenantAsyncExecutor)?
I tried to reproduce the case and it seems that only one is created per tenant.

2

Hey @desp.kaz,

You’ll need to implement your own logic to avoid registering an already registered tenant. I would say that it is not advisable to call the registerTenant twice from different threads in the same time.

Cheers,
Filip

3

Thanks for the answer Filip!
My initial problem was that I may have many kubernetes pods(instances of the app) that should all be aware of the tenants. If only one pod calls the registerTenant method at my listener then I guess that tenantAwareDataSource won’t be updated for the others, right?

4

Yes indeed, the tenantAwareDataSource is a based on a single node implementation. If you have a multi node scenario with a dynamic configuration of those tenants then you should implement something on your own.

Cheers,
Filip

1 Like
5

Thanks!
Just one last question, at my custom implementation the async executor must be created once? Or does it make sense to have one per instance-pod?
I am not entirely sure how this works exactly.

6

It should be one per instance / pod. Otherwise, only one pod will be able to handle async jobs for that particular tenant.

Cheers,
Filip

1 Like