Is it normal to have a bunch of exception logged as WARN in zookeeper test and high cost to analyze history?

[nhahtdh]

When running zookeeper test, I keep getting a bunch of warning logs. Some examples:

INFO  jepsen.util - 170 :invoke :cas    [4 0]
WARN  jepsen.core - Process 170 indeterminate
java.lang.IllegalMonitorStateException

INFO  jepsen.util - :nemesis    :info   :start  "Cut off {:n5 #{:n3 :n4 :n2}, :n1 #{:n3 :n4 :n2}, :n3 #{:n5 :n1}, :n4 #{:n5 :n1}, :n2 #{:n5 :n1}}"
INFO  jepsen.util - 144 :invoke :cas    [4 0]
INFO  jepsen.util - 172 :invoke :cas    [2 3]
WARN  jepsen.core - Process 172 indeterminate
java.lang.IllegalMonitorStateException

INFO  jepsen.control - (iptables -A INPUT -s 192.168.99.222 -j DROP -w)
INFO  jepsen.util - 211 :invoke :read   nil
WARN  jepsen.core - Process 211 indeterminate
org.apache.zookeeper.KeeperException$SessionExpiredException: KeeperErrorCode = Session expired for /jepsen/data

INFO  jepsen.core - Worker 1 done
INFO  org.apache.zookeeper.ClientCnxn - Client session timed out, have not heard from server in 3335ms for sessionid 0x454cc6ebfda0000, closing socket connection and attempting reconnect
WARN  jepsen.core - Process 144 indeterminate
java.lang.IllegalMonitorStateException

Is this normal?

Also, the cost of analyzing the history seems to be pretty high:

INFO  jepsen.core - Analyzing
INFO  knossos.linear - :space 2720 :cost 2.92E+24 :op {:type :ok, :f :read, :value 4, :process 32, :time 17632658846, :index 83}

Is the high cost normal?

[nhahtdh]

Just an extra note. I see hundreds of these exceptions in the log, and I don't see much cas action when exception occurs.

[aphyr]

Yeah, this is normal. If the client crashes, not much you can do. If it bugs you, you can catch exceptions, but Jepsen maps them to info normally. 10^24 is pretty small by knossos standards. If you start getting towards 10^32 you're generally fucked, though. :)