fix httpserver httppipeline status not show error

[suchen-sci]

Fix #438

return status is map -> map[member id]status... Since our egctl does not specific namespace now, we return status from all namespace. this part may need change.

[codecov-commenter]

Codecov Report

Merging #441 (956d1d7) into main (ffd3bdc) will decrease coverage by 0.02%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             main     #441      +/-   ##
==========================================
- Coverage   80.55%   80.52%   -0.03%     
==========================================
  Files          70       70              
  Lines        8094     8154      +60     
==========================================
+ Hits         6520     6566      +46     
- Misses       1218     1229      +11     
- Partials      356      359       +3     

Impacted Files	Coverage Δ
pkg/filter/mock/mock.go	90.42% <0.00%> (-4.66%)	⬇️
pkg/filter/proxy/pool.go	80.00% <0.00%> (-2.66%)	⬇️
pkg/cluster/cluster.go	50.68% <0.00%> (-1.03%)	⬇️
pkg/object/mqttproxy/client.go	77.67% <0.00%> (-0.94%)	⬇️
pkg/object/meshcontroller/spec/spec.go	87.20% <0.00%> (+0.07%)	⬆️
pkg/util/urlclusteranalyzer/urlclusteranalyzer.go	96.10% <0.00%> (+0.38%)	⬆️
pkg/util/jmxtool/agent_controller.go	66.66% <0.00%> (+1.96%)	⬆️
pkg/filter/proxy/request.go	95.31% <0.00%> (+9.59%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ffd3bdc...956d1d7. Read the comment docs.

[samutamm]

I tested the scenario described in issue #438, using curl command curl localhost:2381/apis/v1/status/objects/server-demo. I got more or less the same output with main branch and this.. Would it be possible to list exact steps to reproduce the problem in main branch and then show that these changes fix it?

[suchen-sci]

Hi, @samutamm after start up, you need wait some few seconds before request. easegress will collect status and store it to etcd. We get status from etcd. And if possible, please use egctl object status get server-demo command...

[suchen-sci]

➜  easegress git:(fix-object-status) ✗ ./bin/egctl object status get server-demo
eg-default-name: |
  {}
➜  easegress git:(fix-object-status) ✗ ./bin/egctl object status get server-demo
eg-default-name: |
  {}
➜  easegress git:(fix-object-status) ✗ ./bin/egctl object status get server-demo
eg-default-name: |
  default: |
    health: ""
    state: running
    status:
      count: 0
      m1: 0
      m5: 0
      m15: 0
      errCount: 0
      m1Err: 0
      m5Err: 0
      m15Err: 0
      m1ErrPercent: 0
      m5ErrPercent: 0
      m15ErrPercent: 0
      min: 0
      max: 0
      mean: 0
      p25: 9.999999e+06
      p50: 9.999999e+06
      p75: 9.999999e+06
      p95: 9.999999e+06
      p98: 9.999999e+06
      p99: 9.999999e+06
      p999: 9.999999e+06
      reqSize: 0
      respSize: 0
      codes: {}
    topN: []

here are my results after wait about half minute...

[xxx7xxxx]

Actually, the root cause here is that TrafficController took the control of the lifecycle of HTTPServer/HTTPPipeline, so it's intuitive that the statuses of servers/pipelines are reported by the TrafficController, where it will cause the mislocation between config and status of them.

I'm fine with this quick fix without deep and tidy refactoring. But if there brought more kinds of stuff under TrafficController, we should fix them in a whole policy(TrafficController syncing or move the status to the corresponding location).

[suchen-sci]

@xxx7xxxx @samutamm @localvar I updated the code, previous egctl object status get will return map[member id]map[namespace]status. now it will show map[member id]status. we only show results from the default namespace when use egctl. Is this ok for you?

[xxx7xxxx]

Returning status only from the default namespace is all right.

[samutamm]

LGTM!