web test log to find out why it failed

[moses.moore]

How do I find out why a web test scenario failed? I've only seen records of when it failed, and at what step it failed, but not why that step failed.

I have a web test scenario that fails often, many times a week, but only for one poll period each time. By the time a human can get to it to test with other tools, Zabbix already reports that it's back to normal. None of the other websites on the same machine are suffering; the web test scenario itself is just looking for a '302 Moved Temporarily' response so I know it's not some website app code gone awry.

I could set up a little monitoring script to poll at the same time as Zabbix for the same thing, but if I have to do that then why do I have Zabbix in the first place? So I feel there must be a way for Zabbix to tell me why a web test scenario has failed, not just "yes it failed."

Here's a description of the web test:

Name: cheers
Update Interval: 120
Retries: 1
Agent: Google Chrome 17
Variables:
- {hostname}=[omitted]
Enabled: true
Steps:
- {name: "GET /", url: "http://{hostname}/", follow-redirects: false, timeout: 15, required-status-codes: "301-302"}

And this is the trigger:

{mt:web.test.fail[cheers].last()}<>0

[moses.moore]

So I did set up a bash script to watch the same URL, and carp if there was any change in the output or any problem fetching the URL, polling every 30 seconds while Zabbix polls every 120 seconds.

Zabbix reported a problem with the web scenario at 5:20am while I was sleeping. During the same time, the 30-second polling loop reported that fetching the URL was no different than before.

I really would like to know why Zabbix's web scenario reported a problem when something else running on the same machine polling 4x more frequently did not.

[jan.garaj]

Increase debug level (+log file size also) for web monitoring and you will see the reason in the log file.

http://blog.zabbix.com/zabbix-2-4-features-part-6-runtime-loglevel-changing/3653/

[moses.moore]

Thank you; now I have more info.

cannot process step "GET /" of web scenario "cheers" on host "mt": Timeout was reached: Operation timed out after 15968 milliseconds with 55212 bytes received

I already know from the 30-second loop I had running at the same time that it's not really the remote server timing out, and the extra logging is showing me that many machines among many remote locations are giving me "Operation timed out" and "first network error / connection restored" messages. I'll use what you showed me to investigate further.