Periodically slow response times from /rest_v2/serverInfo

[bill.reynolds]

Hi, 

We're running a 2-node cluster of JasperServer instances, and we needed a consistent URL we could hit to monitor the overall health of each server. From what I could see in the docs, there was no lightweight endpoint like /health we could hit, so we've decided to hit /rest_v2/serverInfo instead.

For the most part, this endpoint works - requests to this URL are often returned in less than 200ms. However, about once an hour we see spikes where it takes upwards of 30 seconds for this URL to return. Has anyone else experienced this?

Is there an alternate URL we can use to monitor the health of the JasperServer? 

We're running JasperServer Professional, version 6.3.0.

Thanks,

-Bill

 

 

 

[vchiem]

If its something to ping to see whether the server is running, then can try a specific call:

http://:/jasperserver[-pro]/rest_v2/serverInfo/version

or another approach may be to send a login request and see if a session id is returned.  

 

[bill.reynolds]

Thanks, though I don't think that endpoint will fix our problem. I did some benchmarking last week against both /serverInfo and /serverInfo/version, and it both cases, both endpoints were subject to the same hourly slowdown. 

I've been inspecting the code with NewRelic, and it both cases we get hung up at the RestLicenseFilter.doFilter() method - I'm guessing this is a servlet filter that handles all REST requests. My visibility into what is going on is limited, but I'm guessing there is a deadlock in that section of code that only manifests itself hourly. As a consequence, it probably will also break if we try to fetch /rest/login, so I'm not going to use that endpoint to monitor health for the application.

In the short term, I'm planning to move our healthcheck to just fetch out /jasperserver-pro/, as that URL returns a 200 response consistently, and I haven't seen it slow down on the hour.