[Ingest Manager] Plugin won't retry failed initial setup #74587

jfsiii · 2020-08-06T19:29:23Z

Steps to reproduce:

Make registry unreachable like (0) in [Ingest Manager] POST /setup crashes Kibana with Unhandled Promise rejection #73997
POST to /setup:
curl -X POST 'http://localhost:5601/api/ingest_manager/setup' -H 'kbn-xsrf: <string>' --user elastic:changeme
observe it fail:
{"statusCode":502,"error":"Bad Gateway","message":"Error connecting to package registry at http://localhost:8080/package/system/0.5.3: request to http://localhost:8080/package/system/0.5.3 failed, reason: connect ECONNREFUSED 127.0.0.1:8080"}
restore registry
repeat 2
observe it fail, like (3)
Other endpoints which hit the registry will succeed, e.g.
curl http://localhost:5601/api/ingest_manager/epm/packages --user elastic:changeme

Movies showing Ingest manager setup retains first result.zip

Root cause
We hold onto the value in setupIngestStatus Promise to prevent simultaneous calls to setup. However, when we reject the promise due to a failure that means we'll never retry since the next await setupIngestStatus will fail

kibana/x-pack/plugins/ingest_manager/server/services/setup.ts

Lines 128 to 134 in beb7b82

    
             onSetupReject(error); 
        
           } 
        
           // be sure to return the promise because it has the resolved/rejected status attached to it 
        
           // otherwise, we effectively return success every time even if there are errors 
        
           // because `return undefined` -> `Promise.resolve(undefined)` in an `async` function 
        
           return setupIngestStatus;

The text was updated successfully, but these errors were encountered:

elasticmachine · 2020-08-06T19:56:31Z

Pinging @elastic/ingest-management (Team:Ingest Management)

* add retries for registry requests. works, afaict. no tests. one TS issue. * Fix TS issue. Add link to node-fetch error docs * Restore some accidentally deleted code. * Add more comments. Remove logging. * Add tests for plugin setup service & handlers * Add tests for Registry retry logic * Extract setup retry logic to separate function/file * Add tests for setup retry logic ``` firstSuccessOrTryAgain ✓ reject/throws is called again & its value returned (18ms) ✓ the first success value is cached (2ms) ``` * More straightforward(?) tests for setup caching * Revert cached setup. Still limit 1 call at a time Terrible tests. Committing & pushing to see if it fixes failures like https://github.com/elastic/kibana/pull/74507/checks?check_run_id=980178887 https://kibana-ci.elastic.co/job/elastic+kibana+pipeline-pull-request/67892/execution/node/663/log/ ``` 07:36:56 └-> "before all" hook 07:36:56 └-> should not allow to enroll an agent with a invalid enrollment 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 │ proc [kibana] error [11:36:56.369] Error: Internal Server Error 07:36:56 │ proc [kibana] at HapiResponseAdapter.toError (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:132:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.toHapiResponse (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:86:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:81:17) 07:36:56 │ proc [kibana] at Router.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/router.js:164:34) 07:36:56 │ proc [kibana] at process._tickCallback (internal/process/next_tick.js:68:7) 07:36:56 │ proc [kibana] log [11:36:56.581] [info][authentication][plugins][security] Authentication attempt failed: [security_exception] missing authentication credentials for REST request [/_security/_authenticate], with { header={ WWW-Authenticate={ 0="ApiKey" & 1="Basic realm=\"security\" charset=\"UTF-8\"" } } } 07:36:56 └- ✓ pass (60ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a invalid enrollment" 07:36:56 └-> should not allow to enroll an agent with a shared id if it already exists 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (111ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a shared id if it already exists " 07:36:56 └-> should not allow to enroll an agent with a version > kibana 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (58ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a version > kibana" 07:36:56 └-> should allow to enroll an agent with a valid enrollment token 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✖ fail: Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should allow to enroll an agent with a valid enrollment token 07:36:56 │ Error: expected 200 "OK", got 500 "Internal Server Error" 07:36:56 │ at Test._assertStatus (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:268:12) 07:36:56 │ at Test._assertFunction (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:283:11) 07:36:56 │ at Test.assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:173:18) 07:36:56 │ at assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:131:12) 07:36:56 │ at /dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:128:5 07:36:56 │ at Test.Request.callback (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:718:3) 07:36:56 │ at parser (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:906:18) 07:36:56 │ at IncomingMessage.res.on (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/parsers/json.js:19:7) 07:36:56 │ at endReadableNT (_stream_readable.js:1145:12) 07:36:56 │ at process._tickCallback (internal/process/next_tick.js:63:19) 07:36:56 │ 07:36:56 │ ``` * New name & tests for one-at-a-time /setup behavior `firstPromiseBlocksAndFufills` for "the first promise created blocks others from being created, then fufills all with that first result" * More (better?) renaming * Fix name in test description * Fix spelling typo. * Remove registry retry code & tests * Use async fn's .catch to avoid unhandled rejection Add explicit `isPending` value instead of overloading role of `status`. Could probably do without it, but it makes the intent more clear. Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

elastic#75372) * add retries for registry requests. works, afaict. no tests. one TS issue. * Fix TS issue. Add link to node-fetch error docs * Restore some accidentally deleted code. * Add more comments. Remove logging. * Add tests for plugin setup service & handlers * Add tests for Registry retry logic * Extract setup retry logic to separate function/file * Add tests for setup retry logic ``` firstSuccessOrTryAgain ✓ reject/throws is called again & its value returned (18ms) ✓ the first success value is cached (2ms) ``` * More straightforward(?) tests for setup caching * Revert cached setup. Still limit 1 call at a time Terrible tests. Committing & pushing to see if it fixes failures like https://github.com/elastic/kibana/pull/74507/checks?check_run_id=980178887 https://kibana-ci.elastic.co/job/elastic+kibana+pipeline-pull-request/67892/execution/node/663/log/ ``` 07:36:56 └-> "before all" hook 07:36:56 └-> should not allow to enroll an agent with a invalid enrollment 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 │ proc [kibana] error [11:36:56.369] Error: Internal Server Error 07:36:56 │ proc [kibana] at HapiResponseAdapter.toError (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:132:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.toHapiResponse (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:86:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:81:17) 07:36:56 │ proc [kibana] at Router.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/router.js:164:34) 07:36:56 │ proc [kibana] at process._tickCallback (internal/process/next_tick.js:68:7) 07:36:56 │ proc [kibana] log [11:36:56.581] [info][authentication][plugins][security] Authentication attempt failed: [security_exception] missing authentication credentials for REST request [/_security/_authenticate], with { header={ WWW-Authenticate={ 0="ApiKey" & 1="Basic realm=\"security\" charset=\"UTF-8\"" } } } 07:36:56 └- ✓ pass (60ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a invalid enrollment" 07:36:56 └-> should not allow to enroll an agent with a shared id if it already exists 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (111ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a shared id if it already exists " 07:36:56 └-> should not allow to enroll an agent with a version > kibana 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (58ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a version > kibana" 07:36:56 └-> should allow to enroll an agent with a valid enrollment token 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✖ fail: Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should allow to enroll an agent with a valid enrollment token 07:36:56 │ Error: expected 200 "OK", got 500 "Internal Server Error" 07:36:56 │ at Test._assertStatus (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:268:12) 07:36:56 │ at Test._assertFunction (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:283:11) 07:36:56 │ at Test.assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:173:18) 07:36:56 │ at assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:131:12) 07:36:56 │ at /dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:128:5 07:36:56 │ at Test.Request.callback (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:718:3) 07:36:56 │ at parser (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:906:18) 07:36:56 │ at IncomingMessage.res.on (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/parsers/json.js:19:7) 07:36:56 │ at endReadableNT (_stream_readable.js:1145:12) 07:36:56 │ at process._tickCallback (internal/process/next_tick.js:63:19) 07:36:56 │ 07:36:56 │ ``` * New name & tests for one-at-a-time /setup behavior `firstPromiseBlocksAndFufills` for "the first promise created blocks others from being created, then fufills all with that first result" * More (better?) renaming * Fix name in test description * Fix spelling typo. * Remove registry retry code & tests * Use async fn's .catch to avoid unhandled rejection Add explicit `isPending` value instead of overloading role of `status`. Could probably do without it, but it makes the intent more clear. Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

elastic#75372) * add retries for registry requests. works, afaict. no tests. one TS issue. * Fix TS issue. Add link to node-fetch error docs * Restore some accidentally deleted code. * Add more comments. Remove logging. * Add tests for plugin setup service & handlers * Add tests for Registry retry logic * Extract setup retry logic to separate function/file * Add tests for setup retry logic ``` firstSuccessOrTryAgain ✓ reject/throws is called again & its value returned (18ms) ✓ the first success value is cached (2ms) ``` * More straightforward(?) tests for setup caching * Revert cached setup. Still limit 1 call at a time Terrible tests. Committing & pushing to see if it fixes failures like https://github.com/elastic/kibana/pull/74507/checks?check_run_id=980178887 https://kibana-ci.elastic.co/job/elastic+kibana+pipeline-pull-request/67892/execution/node/663/log/ ``` 07:36:56 └-> "before all" hook 07:36:56 └-> should not allow to enroll an agent with a invalid enrollment 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 │ proc [kibana] error [11:36:56.369] Error: Internal Server Error 07:36:56 │ proc [kibana] at HapiResponseAdapter.toError (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:132:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.toHapiResponse (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:86:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:81:17) 07:36:56 │ proc [kibana] at Router.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/router.js:164:34) 07:36:56 │ proc [kibana] at process._tickCallback (internal/process/next_tick.js:68:7) 07:36:56 │ proc [kibana] log [11:36:56.581] [info][authentication][plugins][security] Authentication attempt failed: [security_exception] missing authentication credentials for REST request [/_security/_authenticate], with { header={ WWW-Authenticate={ 0="ApiKey" & 1="Basic realm=\"security\" charset=\"UTF-8\"" } } } 07:36:56 └- ✓ pass (60ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a invalid enrollment" 07:36:56 └-> should not allow to enroll an agent with a shared id if it already exists 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (111ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a shared id if it already exists " 07:36:56 └-> should not allow to enroll an agent with a version > kibana 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (58ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a version > kibana" 07:36:56 └-> should allow to enroll an agent with a valid enrollment token 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✖ fail: Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should allow to enroll an agent with a valid enrollment token 07:36:56 │ Error: expected 200 "OK", got 500 "Internal Server Error" 07:36:56 │ at Test._assertStatus (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:268:12) 07:36:56 │ at Test._assertFunction (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:283:11) 07:36:56 │ at Test.assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:173:18) 07:36:56 │ at assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:131:12) 07:36:56 │ at /dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:128:5 07:36:56 │ at Test.Request.callback (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:718:3) 07:36:56 │ at parser (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:906:18) 07:36:56 │ at IncomingMessage.res.on (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/parsers/json.js:19:7) 07:36:56 │ at endReadableNT (_stream_readable.js:1145:12) 07:36:56 │ at process._tickCallback (internal/process/next_tick.js:63:19) 07:36:56 │ 07:36:56 │ ``` * New name & tests for one-at-a-time /setup behavior `firstPromiseBlocksAndFufills` for "the first promise created blocks others from being created, then fufills all with that first result" * More (better?) renaming * Fix name in test description * Fix spelling typo. * Remove registry retry code & tests * Use async fn's .catch to avoid unhandled rejection Add explicit `isPending` value instead of overloading role of `status`. Could probably do without it, but it makes the intent more clear. Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> # Conflicts: # x-pack/plugins/ingest_manager/server/services/setup.ts

…) (#75583) * add retries for registry requests. works, afaict. no tests. one TS issue. * Fix TS issue. Add link to node-fetch error docs * Restore some accidentally deleted code. * Add more comments. Remove logging. * Add tests for plugin setup service & handlers * Add tests for Registry retry logic * Extract setup retry logic to separate function/file * Add tests for setup retry logic ``` firstSuccessOrTryAgain ✓ reject/throws is called again & its value returned (18ms) ✓ the first success value is cached (2ms) ``` * More straightforward(?) tests for setup caching * Revert cached setup. Still limit 1 call at a time Terrible tests. Committing & pushing to see if it fixes failures like https://github.com/elastic/kibana/pull/74507/checks?check_run_id=980178887 https://kibana-ci.elastic.co/job/elastic+kibana+pipeline-pull-request/67892/execution/node/663/log/ ``` 07:36:56 └-> "before all" hook 07:36:56 └-> should not allow to enroll an agent with a invalid enrollment 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 │ proc [kibana] error [11:36:56.369] Error: Internal Server Error 07:36:56 │ proc [kibana] at HapiResponseAdapter.toError (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:132:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.toHapiResponse (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:86:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:81:17) 07:36:56 │ proc [kibana] at Router.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/router.js:164:34) 07:36:56 │ proc [kibana] at process._tickCallback (internal/process/next_tick.js:68:7) 07:36:56 │ proc [kibana] log [11:36:56.581] [info][authentication][plugins][security] Authentication attempt failed: [security_exception] missing authentication credentials for REST request [/_security/_authenticate], with { header={ WWW-Authenticate={ 0="ApiKey" & 1="Basic realm=\"security\" charset=\"UTF-8\"" } } } 07:36:56 └- ✓ pass (60ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a invalid enrollment" 07:36:56 └-> should not allow to enroll an agent with a shared id if it already exists 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (111ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a shared id if it already exists " 07:36:56 └-> should not allow to enroll an agent with a version > kibana 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (58ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a version > kibana" 07:36:56 └-> should allow to enroll an agent with a valid enrollment token 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✖ fail: Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should allow to enroll an agent with a valid enrollment token 07:36:56 │ Error: expected 200 "OK", got 500 "Internal Server Error" 07:36:56 │ at Test._assertStatus (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:268:12) 07:36:56 │ at Test._assertFunction (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:283:11) 07:36:56 │ at Test.assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:173:18) 07:36:56 │ at assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:131:12) 07:36:56 │ at /dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:128:5 07:36:56 │ at Test.Request.callback (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:718:3) 07:36:56 │ at parser (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:906:18) 07:36:56 │ at IncomingMessage.res.on (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/parsers/json.js:19:7) 07:36:56 │ at endReadableNT (_stream_readable.js:1145:12) 07:36:56 │ at process._tickCallback (internal/process/next_tick.js:63:19) 07:36:56 │ 07:36:56 │ ``` * New name & tests for one-at-a-time /setup behavior `firstPromiseBlocksAndFufills` for "the first promise created blocks others from being created, then fufills all with that first result" * More (better?) renaming * Fix name in test description * Fix spelling typo. * Remove registry retry code & tests * Use async fn's .catch to avoid unhandled rejection Add explicit `isPending` value instead of overloading role of `status`. Could probably do without it, but it makes the intent more clear. Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

elastic#75372) * add retries for registry requests. works, afaict. no tests. one TS issue. * Fix TS issue. Add link to node-fetch error docs * Restore some accidentally deleted code. * Add more comments. Remove logging. * Add tests for plugin setup service & handlers * Add tests for Registry retry logic * Extract setup retry logic to separate function/file * Add tests for setup retry logic ``` firstSuccessOrTryAgain ✓ reject/throws is called again & its value returned (18ms) ✓ the first success value is cached (2ms) ``` * More straightforward(?) tests for setup caching * Revert cached setup. Still limit 1 call at a time Terrible tests. Committing & pushing to see if it fixes failures like https://github.com/elastic/kibana/pull/74507/checks?check_run_id=980178887 https://kibana-ci.elastic.co/job/elastic+kibana+pipeline-pull-request/67892/execution/node/663/log/ ``` 07:36:56 └-> "before all" hook 07:36:56 └-> should not allow to enroll an agent with a invalid enrollment 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 │ proc [kibana] error [11:36:56.369] Error: Internal Server Error 07:36:56 │ proc [kibana] at HapiResponseAdapter.toError (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:132:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.toHapiResponse (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:86:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:81:17) 07:36:56 │ proc [kibana] at Router.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/router.js:164:34) 07:36:56 │ proc [kibana] at process._tickCallback (internal/process/next_tick.js:68:7) 07:36:56 │ proc [kibana] log [11:36:56.581] [info][authentication][plugins][security] Authentication attempt failed: [security_exception] missing authentication credentials for REST request [/_security/_authenticate], with { header={ WWW-Authenticate={ 0="ApiKey" & 1="Basic realm=\"security\" charset=\"UTF-8\"" } } } 07:36:56 └- ✓ pass (60ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a invalid enrollment" 07:36:56 └-> should not allow to enroll an agent with a shared id if it already exists 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (111ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a shared id if it already exists " 07:36:56 └-> should not allow to enroll an agent with a version > kibana 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (58ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a version > kibana" 07:36:56 └-> should allow to enroll an agent with a valid enrollment token 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✖ fail: Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should allow to enroll an agent with a valid enrollment token 07:36:56 │ Error: expected 200 "OK", got 500 "Internal Server Error" 07:36:56 │ at Test._assertStatus (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:268:12) 07:36:56 │ at Test._assertFunction (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:283:11) 07:36:56 │ at Test.assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:173:18) 07:36:56 │ at assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:131:12) 07:36:56 │ at /dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:128:5 07:36:56 │ at Test.Request.callback (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:718:3) 07:36:56 │ at parser (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:906:18) 07:36:56 │ at IncomingMessage.res.on (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/parsers/json.js:19:7) 07:36:56 │ at endReadableNT (_stream_readable.js:1145:12) 07:36:56 │ at process._tickCallback (internal/process/next_tick.js:63:19) 07:36:56 │ 07:36:56 │ ``` * New name & tests for one-at-a-time /setup behavior `firstPromiseBlocksAndFufills` for "the first promise created blocks others from being created, then fufills all with that first result" * More (better?) renaming * Fix name in test description * Fix spelling typo. * Remove registry retry code & tests * Use async fn's .catch to avoid unhandled rejection Add explicit `isPending` value instead of overloading role of `status`. Could probably do without it, but it makes the intent more clear. Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

…) (#75587) * add retries for registry requests. works, afaict. no tests. one TS issue. * Fix TS issue. Add link to node-fetch error docs * Restore some accidentally deleted code. * Add more comments. Remove logging. * Add tests for plugin setup service & handlers * Add tests for Registry retry logic * Extract setup retry logic to separate function/file * Add tests for setup retry logic ``` firstSuccessOrTryAgain ✓ reject/throws is called again & its value returned (18ms) ✓ the first success value is cached (2ms) ``` * More straightforward(?) tests for setup caching * Revert cached setup. Still limit 1 call at a time Terrible tests. Committing & pushing to see if it fixes failures like https://github.com/elastic/kibana/pull/74507/checks?check_run_id=980178887 https://kibana-ci.elastic.co/job/elastic+kibana+pipeline-pull-request/67892/execution/node/663/log/ ``` 07:36:56 └-> "before all" hook 07:36:56 └-> should not allow to enroll an agent with a invalid enrollment 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 │ proc [kibana] error [11:36:56.369] Error: Internal Server Error 07:36:56 │ proc [kibana] at HapiResponseAdapter.toError (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:132:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.toHapiResponse (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:86:19) 07:36:56 │ proc [kibana] at HapiResponseAdapter.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/response_adapter.js:81:17) 07:36:56 │ proc [kibana] at Router.handle (/dev/shm/workspace/parallel/5/kibana/build/kibana-build-xpack/src/core/server/http/router/router.js:164:34) 07:36:56 │ proc [kibana] at process._tickCallback (internal/process/next_tick.js:68:7) 07:36:56 │ proc [kibana] log [11:36:56.581] [info][authentication][plugins][security] Authentication attempt failed: [security_exception] missing authentication credentials for REST request [/_security/_authenticate], with { header={ WWW-Authenticate={ 0="ApiKey" & 1="Basic realm=\"security\" charset=\"UTF-8\"" } } } 07:36:56 └- ✓ pass (60ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a invalid enrollment" 07:36:56 └-> should not allow to enroll an agent with a shared id if it already exists 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (111ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a shared id if it already exists " 07:36:56 └-> should not allow to enroll an agent with a version > kibana 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✓ pass (58ms) "Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should not allow to enroll an agent with a version > kibana" 07:36:56 └-> should allow to enroll an agent with a valid enrollment token 07:36:56 └-> "before each" hook: global before each 07:36:56 └-> "before each" hook: beforeSetupWithDockerRegistry 07:36:56 └- ✖ fail: Ingest Manager Endpoints Fleet Endpoints fleet_agents_enroll should allow to enroll an agent with a valid enrollment token 07:36:56 │ Error: expected 200 "OK", got 500 "Internal Server Error" 07:36:56 │ at Test._assertStatus (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:268:12) 07:36:56 │ at Test._assertFunction (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:283:11) 07:36:56 │ at Test.assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:173:18) 07:36:56 │ at assert (/dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:131:12) 07:36:56 │ at /dev/shm/workspace/kibana/node_modules/supertest/lib/test.js:128:5 07:36:56 │ at Test.Request.callback (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:718:3) 07:36:56 │ at parser (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/index.js:906:18) 07:36:56 │ at IncomingMessage.res.on (/dev/shm/workspace/kibana/node_modules/superagent/lib/node/parsers/json.js:19:7) 07:36:56 │ at endReadableNT (_stream_readable.js:1145:12) 07:36:56 │ at process._tickCallback (internal/process/next_tick.js:63:19) 07:36:56 │ 07:36:56 │ ``` * New name & tests for one-at-a-time /setup behavior `firstPromiseBlocksAndFufills` for "the first promise created blocks others from being created, then fufills all with that first result" * More (better?) renaming * Fix name in test description * Fix spelling typo. * Remove registry retry code & tests * Use async fn's .catch to avoid unhandled rejection Add explicit `isPending` value instead of overloading role of `status`. Could probably do without it, but it makes the intent more clear. Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> # Conflicts: # x-pack/plugins/ingest_manager/server/services/setup.ts Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

jfsiii self-assigned this Aug 6, 2020

thomasneirynck added Team:Fleet Team label for Observability Data Collection Fleet team triage_needed labels Aug 6, 2020

This was referenced Aug 6, 2020

[Ingest Manager] Retry for failed request to Registry #74598

Closed

[Ingest Manager] Add retries for registry requests. fixes #74598 #74507

Merged

jfsiii removed the triage_needed label Aug 6, 2020

jfsiii mentioned this issue Aug 19, 2020

[Ingest Manager] Don't retain POST /setup results. fixes #74587 #75372

Merged

1 task

jfsiii closed this as completed in #75372 Aug 20, 2020

jfsiii mentioned this issue Aug 20, 2020

[7.x] [Ingest Manager] Don't retain POST /setup results. fixes #74587 (#75372) #75583

Merged

jfsiii mentioned this issue Aug 20, 2020

[7.9] [Ingest Manager] Don't retain POST /setup results. fixes #74587 (#75372) #75587

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Ingest Manager] Plugin won't retry failed initial setup #74587

[Ingest Manager] Plugin won't retry failed initial setup #74587

jfsiii commented Aug 6, 2020 •

edited

Loading

elasticmachine commented Aug 6, 2020

[Ingest Manager] Plugin won't retry failed initial setup #74587

[Ingest Manager] Plugin won't retry failed initial setup #74587

Comments

jfsiii commented Aug 6, 2020 • edited Loading

elasticmachine commented Aug 6, 2020

jfsiii commented Aug 6, 2020 •

edited

Loading