Our director wants to know why our Mailboxd: Response Time by Client Protocol average is so much higher on our master server than our other servers.
Looking at the soap.csv times and the values in mailboxd.csv, I cannot figure out how this average is calculated, or why it is so much higher than the actual soap requests.
I've looked through the code for zmlogger and all the zmstats-*. I tried to fuser and lsof mailboxd.csv but I can't find what process is writting to this every minute:
while [ 1 ]
> fuser mailboxd.csv
> sleep 1
Cannot stat file /proc/21095/fd/1101: No such file or directory
Values in mailboxd.csv:
soap_count Number of SOAP requests received
soap_ms_avg Average processing time (ms) of SOAP requests
Could anyone point me to the code that calculates this average or where it is getting the times to calculate the average? I know that users are not seeing the 4,000millisecond average the charts are reporting, but I need to prove it.