I actually haven’t tried, but I am quite sure that is you point your logs to a shared location ( in my case hdfs), all node will write to the same place (in my example I pointed them to a local dir but you could use a shared location like S3). In that case, you also will need to configure the apnder to support async writes so the log file don’t get corrupted by all nodes writing at the same time.

Please try and let me know.