1

We have a Zabbix server that reports on Tomcat's errorCount, from the GlobalRequestProcessor. I'm trying to figure out exactly what gets counted in this errorCount. Is it any request to Tomcat the results in an error? If so, how are these measured? Any request that results in an HTTP error response code (4xx/5xx)? Are there other conditions that would also affect errorCount that would not affect the HTTP response code?

Basically, if someone sees a chart showing that the error rate is increasing, what should they look at to see where the errors are coming from: Application logs? Tomcat logs? Apache web logs? Something else?

FrustratedWithFormsDesigner
  • 123
  • 1
  • 7

1 Answers1

1

Your hypothesis is correct: as the comment in the org.apache.coyote.RequestInfo class says:

// number of response codes >= 400
private int errorCount;

the errorCount is incremented for each request that results in a 4xx/5xx status code and nothing more. SSL handshake errors will not get counted, as reported in this question, since they don't create a request.

The 4xx/5xx errors will certainly leave a trace in the access log (localhost_access.log in the default config) and if the error was caused by an exception, it will probably find its way to catalina.<date>.log.

I don't know, which role plays Apache2 in your installation, but if it serves as reverse proxy it will mirror Tomcat's access log with some errors of its own.

Piotr P. Karwasz
  • 5,748
  • 2
  • 11
  • 21