Recently our webapp + EventStore caused out disk errors, because the
log went crazy with this type of message:
eventstore.ServerErrorException: Log record at actual pos 782205 has
too large length: 543516513 bytes, while limit is 16777216 bytes. In
chunk #0-0 (chunk-000000.000000).
It kept doing that until we were out of disk. My guess is this is
because I forcibly replaced the read model database with one I had
generated on my computer, without then resetting EventStore. So, on
startup the app reads the last known position from the db, and does a
subscription from that point. But because that position is "weird" to
that instance of EventStore, it all goes downhill fast.
Does anyone have good patterns for clients to detect this on startup,
to avoid running out of disk because of the log? Has anyone else run
into similar issues?