SSL connect error

[josh1]

Hi
I set a web scenario for an url(https://192.168.1.1/test).
It worked well but sometimes failed(7-8 times a day) with a message "SSL connect error".
The red line in the attached pix mean a failed case.
I tried to poll url by curl directly as comparison and it worked well all the time.
Any advice will be appreciated


**error message from zabbix log
23842:20140915:170259.432 cannot process step "test_443" of web scenario "test_443" on host "192.168.1.1(Linux)": SSL connect error

[Aleksejs]

Hi!
Do you have access to log files of web server that you are monitoring?
Are there maybe corresponding status:408 log entries?
It would mean that the SSL handshake timed out.

[josh1]

Hi
I saw some error messages in the zabbix log and it looks like following.
Is this the log you meant?
Thanks

/var/log/zabbix/zabbix_server.log
23842:20140917:024701.349 cannot process step "test_443" of web scenario "test_443" on host "192.168.1.1(Linux)": SSL connect error

[Aleksejs]

No, I meant webserver that you are monitoring logs.
In case of apache look for line like (if your zabbix server IP address is 192.168.1.123):
[18/Sep/2014:00:02:44 +0300] 192.168.1.123 TLSv1.1 ECDHE-RSA-AES256-SHA "-" "-" "-" status:408 byte:- time:0 "-" "-"

On numerous occasions I've seen SSL handshake timing out.
Meaning that web server closes connection if client (in this case zabbix server) does not end SSL handshake in the required time.

The Web server (running the Web site) thinks that there has been too long an interval of time between 1) the establishment of an IP connection (socket) between the client (e.g. your Web browser or our CheckUpDown robot) and the server and 2) the receipt of any data on that socket, so the server has dropped the connection. The socket connection has actually been lost - the Web server has 'timed out' on that particular socket connection. The request from the client must be repeated - in a timely manner.

Fixing 408 errors - general

408 errors are often difficult to resolve. They typically involve one-off variations in system workload or operations.

If you see persistent 408 errors, the first thing to consider is the workload on the Web server - particularly around the time the 408 errors were generated. If this is light, then you also need to consider workload on the client system. If the computer systems on both ends of the socket connection seem to be running normally, then temporary Internet surges may be to blame.

Hope that it helps. You can try fiddling with your web servers settings if it is the case.

[josh1]

Hi
I am running a Tomcat6 service.
I can't see the error about E408 or any other error about http.
Would have a Tomcat experience of debuging?

Thanks for your help

[Aleksejs]

I'm more of a WebSphere guy, but I think, access logs are disabled by default on tomcat.
http://blog.techstacks.com/2008/08/t...s-logging.html