When my server has a lot of people on, /spark heapdump crashes the server and I never get to take the heapdump to see what is using the most memory.

Here is pastebin of a heapdump I did with only 8 players on (although this one didnt crash, it froze for about 30 seconds before giving a stack trace. Usually they crash on my main servers)

https://pastebin.com/iqbVgykR

So your issue is that your server crashes, but you gave us info about the time it didn't crash. You aren't helping very much... Creation of heap dumps take a lot of time (depending on how much memory you have), this is not a spark-specific issue.

Are you sure the server actually crashed when you thought it crashed? It not responding for a few minutes might be normal, but again, it depends on how much memory you have allocated.

spark just asks the VM to create a heapdump - it doesn't actually do anything special to produce the output itself.

If you server is crashing as a result, there's not much that can be done. You might have better luck using another tool (not spark) - something from the command line - to generate the help dump instead.