Open
Description
I'm just discovering a state at UCSD where the aggregator record
process has crashed due to some type error, but new data coming in continues to get added to the data queue. Since the aggregator version is fairly old, it's possible that the type error may be fixed, but I think the behavior of data being added to the queue when record
is not running probably still exists.
I think we probably want this behavior so we don't lose data on crash, but we should probably be setting a max queue size to limit memory usage.