[Uptime] While running functional tests, undefined exception

[shahzad31]

While running uptime functional tests, sometimes this exception get's throwned.

   │ proc [kibana] TypeError: Cannot read property 'monitors' of undefined
   │ proc [kibana]     at findPotentialMatches (/Users/shahzad/elastic/kibana/x-pack/plugins/uptime/server/lib/requests/search/find_potential_matches.ts:50:43)
   │ proc [kibana]     at process._tickCallback (internal/process/next_tick.js:68:7)
   │ proc [kibana] server    log   [13:12:38.338] [error][http] TypeError: Cannot read property 'counts' of undefined
   │ proc [kibana]     at statusCount (/Users/shahzad/elastic/kibana/x-pack/plugins/uptime/server/lib/requests/get_snapshot_counts.ts:58:27)
   │ proc [kibana]     at process._tickCallback (internal/process/next_tick.js:68:7)
   │ proc [kibana] server   error  [13:12:38.166]  Error: Internal Server Error
   │ proc [kibana]     at HapiResponseAdapter.toInternalError (/Users/shahzad/elastic/kibana/src/core/server/http/router/response_adapter.ts:56:19)
   │ proc [kibana]     at Router.handle (/Users/shahzad/elastic/kibana/src/core/server/http/router/router.ts:268:34)
   │ proc [kibana]     at process._tickCallback (internal/process/next_tick.js:68:7)

[elasticmachine]

Pinging @elastic/uptime (Team:uptime)

[justinkambic]

I have seen this as well. I think it's not really a product of the tests having a problem, but more of the server not handling undefined keys on result objects.

The solution for the thrown exception is probably to make the server code more resilient, and just return an empty set.

The actual cause of the empty set is (I think) that the index pattern the app is trying to use does not exist. We should investigate by logging the parameters the server is sending to Elasticsearch just before the exception is thrown to make sure of this, and return a nice message like "Uptime index pattern does not exist".