The CNDC AutoDRM system was originally designed primarily to supply waveform data from one or more stations for relatively short time segments typically coinciding with a specific earthquake. With the advent of faster computers and a fully online archive, many of the earlier restrictions on number of stations and time spans have been relaxed, but there are still limits.
Some "power users" employ scripts to generate a batch of autodrm e-mail requests in one go. While very efficient from the individual user's perspective, processing a large number of requests from a single user in sequence effectively ties up the CNDC autodrm service and locks out other users. On a number of occasions in the past, when scripts went awry and clogged the system, we have had to shut down autodrm processing and flush all unprocessed requests from the input queue in order to restore normal operations. Clearly, this makes nobody happy.
One strategy that works well when scripts are used to generate batch requests is to incorporate a "sleep" of 30-60 seconds between individual requests. This allows others fair access, while taking little more work or time.