Arvados

A user reported that they cannot change the page size (number of items in the data table) on the "All Processes". I've confirmed the issue and realized that workbench was continuously making requests like:

GET https://zzzz.example.com/arvados/v1/container_requests?limit=100&offset=0&order=created_at+desc,created_at+desc&filters=[[%22name%22,%22ilike%22,%22%25%25%22],[%22requesting_container_uuid%22,%22%3D%22,null]]&select=[%22command%22,+%22container_count_max%22,+%22container_count%22,+%22container_image%22,+%22container_uuid%22,+%22created_at%22,+%22cumulative_cost%22,+%22cwd%22,+%22description%22,+%22environment%22,+%22etag%22,+%22expires_at%22,+%22filters%22,+%22href%22,+%22kind%22,+%22log_uuid%22,+%22modified_at%22,+%22modified_by_client_uuid%22,+%22modified_by_user_uuid%22,+%22name%22,+%22output_name%22,+%22output_path%22,+%22output_properties%22,+%22output_storage_classes%22,+%22output_ttl%22,+%22output_uuid%22,+%22owner_uuid%22,+%22priority%22,+%22properties%22,+%22requesting_container_uuid%22,+%22runtime_constraints%22,+%22scheduling_parameters%22,+%22state%22,+%22use_existing%22,+%22uuid%22]
POST https://zzzz.example.com/arvados/v1/containers

I assume this is workbench refreshing the data table contents, possibly based on some event received from websockets.
Changing pages works fine, but changing the page size requires several attempts and eventually works. This is happening on a cluster with 100+ running containers, I've checked other less busy cluster and haven't seen the issue happening.