Currently investigating high load on one of our databases and trying to track down the culprit / look for any optomizations.
Using tcpdump and pt-query-digest one query which always appears at the top of the reports wether sorted by max total execution time or frequency of calls is the "ADMIN PING" command as seen below:
Profile
Rank Query ID Response time Calls R/Call Apdx V/M Item
==== ================== ============= ====== ====== ==== ===== =========
1 0x04FE01C5B31FD305 10.8718 7.7% 140851 0.0001 1.00 0.03 ADMIN PING
From what I can gather from the documentaion this is just to see if the server is still alive, though just running your statement and then checking for errors would answer the question if the server is responding or not.
I haven't tracked down the ADMIN PING to the specific application yet but i suspect it's to do with perl DBI (We have perl/java jdbc/C/ruby apps accessing the same server)
My specific question is does the number of ADMIN PING calls seem excessive (the line above is from a report of a 15 second tcpdump)?
Secondly can the ADMIN PING be disabled without issue and are there any docs that show how to enable/disable the feature for Perl DBI and Java JDBC ?
