EMI ICP connection data API response delays – 10 Mar 2019

  • This discussion is locked, please start a new discussion.
  • 617 Views
  • Last post 11 March 2019
  • Discussion is solved
funcsol posted this 10 March 2019

Hi there, 

 

ICP connection data API is apparently taking a signficant time to return for each polling attempt (anecdotally about a minute). Is there a currently known performance issue in place? 

We have also taken the step of checking with Jade & there is not any performance issues from their side. 

 

regards,

  • Liked by
  • michael@glimp.co.nz
Order by: Standard | Newest | Votes
michael@glimp.co.nz posted this 10 March 2019

Same issue here. 

funcsol posted this 10 March 2019

As at now(ish) the response times appear to be normal. 

 

Regards. 

Phil Bishop posted this 10 March 2019

Our logs show that all recent requests - since about 3pm on Sunday - have been achieved in times of less than one second. However, prior to that there were some slow requests taking up to a minute.

We're still finessing the logging database solution for this API. Subject to some final testing that is occuring at the minute, we're confident that we've landed on a durable and effective solution. We'll continue to monitor performance.

 

Thanks,

Phil

michael@glimp.co.nz posted this 10 March 2019

Hi everyone,

Looks like we are getting slow responses again at well over a minute.

 

  • Liked by
  • funcsol
funcsol posted this 11 March 2019

Performance is still variable, currently very slow. Im not sure what you are logging to show that the response is sub 1 second, I find that test queries are now taking longer than 30 seconds minimum. 

funcsol posted this 11 March 2019

Per the health graph of the API... there are large & current lags in response time. 

API Health Graph 

Regards, 

SimonT posted this 11 March 2019

We're also seeing a number of time-out errors - the publc site is also taking a long time to respond to queries

Rachel Langford (PKS) posted this 11 March 2019

Performance should be resuming to regular levels; the logging databases were still causing some issues after noon today, causing the slowdowns, and this has been resolved.

  • Liked by
  • funcsol
  • SimonT

This discussion is locked, please start a new discussion.