nens / parramatta-dashboard

Dashboard web-app for the City of Parramatta
https://parramatta.lizard.net/clients/parramatta/
GNU General Public License v3.0
0 stars 0 forks source link

Graphs in Parramatta Trainings Dashboard are broken. #241

Closed joeriverheijden closed 3 years ago

joeriverheijden commented 3 years ago

Graphs in the parramatta trainings dashboard retrieve their data from the client_config. There seems to be an issues that the hard-coded data in the client_config is not visualized as graphs in the dashboard. The graphs in the dashboard remain blank.

See screen dump: image.png

joeriverheijden commented 3 years ago

@remcogerlich the ticket is now made in the Lizard Client repo (couldn't find the parramatta-tile repo.

joeriverheijden commented 3 years ago

@Derryrover @lexvand dthis is a high prio bug. Parramatta trainingsdashboards are currently not usable!!!

Derryrover commented 3 years ago

@lexvand @remcogerlich Hi Lex and Remco,

Joeri asked me to pick this up with high prio!

kr Tom

Derryrover commented 3 years ago

@joeriverheijden

Hi Joeri,

Do you happen to know:

I will need to find which change could have caused this. But there are no changes other then these since may: https://github.com/nens/parramatta-dashboard/compare/v0.3.34...master Therefore it is probably a configuration issue. I will continue investigation.

kr Tom

joeriverheijden commented 3 years ago

This is reported a fews days back

This is the first time that it happens.

As far as I know the configuration hasn't changed....

Op di 29 sep. 2020 om 14:43 schreef Tom de Boer notifications@github.com:

@joeriverheijden https://github.com/joeriverheijden

Hi Joeri,

Do you happen to know:

  • When was this first reported?
  • Is this the first time happening?

I will need to find which change could have caused this. But there are no changes other then these since may: v0.3.34...master https://github.com/nens/parramatta-dashboard/compare/v0.3.34...master Therefore it is probably a configuration issue. I will continue investigation.

kr Tom

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/nens/parramatta-dashboard/issues/241#issuecomment-700677124, or unsubscribe https://github.com/notifications/unsubscribe-auth/AELLJSCQLUOUBPBXMDNKXBDSIHI7XANCNFSM4R4MQUNQ .

--

Met vriendelijke groet / Kind regards,

Joeri Verheijden

Product Manager Nelen & Schuurmans

+31 (0)30 233 0200 | +31 (0)6 21586741 www.nelen-schuurmans.nl Lizard https://www.lizard.net/ 3Di https://www.3diwatermanagement.com/ Zakkendragershof 34-44, 3511AE, Utrecht [image: Logo N&S] [image: twitter] https://twitter.com/nelenschuurmans [image: linkedin] https://www.linkedin.com/company/245070/

Derryrover commented 3 years ago

I see the following raster api call returning a 404: https://parramatta.lizard.net/api/v3/rasters/b1153ca/?format=json Or whole uuid: https://parramatta.lizard.net/api/v3/rasters/b1153ca3-1268-44f2-9df1-ef0666ac4499/?format=json

Could this possible be the cause? Which raster was this? Can we still retrieve it?

lisannewapstra commented 3 years ago

Hi Tom, I had the same and tested with another raster (the Rain Forecast for Parramatta operational dashboard) in trainingscenario1 and it removes the 404 error, but the tiles do not work.

Derryrover commented 3 years ago

I suspect now as follows, but tests need to verify if this is true:

The following timeseries is used as "fakedata": 49969142-7890-4177-8cb3-7b920715ac7a baebfd87-8283-43e6-afd3-6bb7513ab578 5eb1fad7-180d-44d5-a83b-74105760cddc a85cbac3-34de-4d1c-8799-8f87475c1543 baebfd87-8283-43e6-afd3-6bb7513ab578

Anyway, an api call to retrieve metadata for this fake data is still done (probably a bug). ->This api call does not return the metadata (has it been deleted?). -> The metadata does then not end up in the redux store -> the metadata is then not found by the plotly tile -> the whole plotly graph is therefore not drawn

The long term fix should be that metadata is not retrieved from backend but used from the fakedata in the clientconfig (it seems to be already there). A short term fix may be restoring the missing timeseries.

Derryrover commented 3 years ago

@remcogerlich

Hi Remco, See my above observation. Did you for making the training dashboards to work change in the lizard-api-client repo how timeseries metadata was retrieved? I can only find the following change: https://github.com/nens/lizard-api-client/pull/26

Or is the fakedata retrieved in the parramatta app itself ?

kr Tom

remcogerlich commented 3 years ago

Did the requests change from v3 to v4? I seem to recall that v4 split up the metadata and data requests, or something like that.

Derryrover commented 3 years ago

Did the requests change from v3 to v4?

No they are still v3

Derryrover commented 3 years ago

@remcogerlich I think I have a partial solution in this pr. https://github.com/nens/parramatta-dashboard/pull/243 You can proxy to prod and use this url to test: http://localhost:3000/floodsmart/trainingscenario1/

Derryrover commented 3 years ago

@joeriverheijden @lisannewapstra @remcogerlich

The following data is missing in the backend. This probably causes the problems. Most urgent is the missing raster. The missing timeseries can also be fixed with a patch in the pr.

raster (Urgent): b1153ca3-1268-44f2-9df1-ef0666ac4499

timeseries (not urgent, fix is available in pr): 49969142-7890-4177-8cb3-7b920715ac7a baebfd87-8283-43e6-afd3-6bb7513ab578 5eb1fad7-180d-44d5-a83b-74105760cddc a85cbac3-34de-4d1c-8799-8f87475c1543 baebfd87-8283-43e6-afd3-6bb7513ab578

joeriverheijden commented 3 years ago

I've asked Jeroen to figure out if these rasters are necessary in the config file. (my feeling is that we can delete them from the config file as they are not visualised in the dashboard)

Jeroen will update me asap.

Op do 1 okt. 2020 om 11:41 schreef Tom de Boer notifications@github.com:

@joeriverheijden https://github.com/joeriverheijden @lisannewapstra https://github.com/lisannewapstra @remcogerlich https://github.com/remcogerlich

The following data is missing in the backend. This probably causes the problems. Most urgent is the missing raster. The missing timeseries can also be fixed with a patch in the pr.

raster (Urgent): b1153ca3-1268-44f2-9df1-ef0666ac4499

timeseries (not urgent, fix is available in pr): 49969142-7890-4177-8cb3-7b920715ac7a baebfd87-8283-43e6-afd3-6bb7513ab578 5eb1fad7-180d-44d5-a83b-74105760cddc a85cbac3-34de-4d1c-8799-8f87475c1543 baebfd87-8283-43e6-afd3-6bb7513ab578

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/nens/parramatta-dashboard/issues/241#issuecomment-702018095, or unsubscribe https://github.com/notifications/unsubscribe-auth/AELLJSARXCWI2IZDWGJO5RDSIRFEHANCNFSM4R4MQUNQ .

--

Met vriendelijke groet / Kind regards,

Joeri Verheijden

Product Manager Nelen & Schuurmans

+31 (0)30 233 0200 | +31 (0)6 21586741 www.nelen-schuurmans.nl Lizard https://www.lizard.net/ 3Di https://www.3diwatermanagement.com/ Zakkendragershof 34-44, 3511AE, Utrecht [image: Logo N&S] [image: twitter] https://twitter.com/nelenschuurmans [image: linkedin] https://www.linkedin.com/company/245070/

remcogerlich commented 3 years ago

The rasters aren't drawn on maps I think, but there are timeseries from the rasters drawn in charts. And for that the raster metadata is necessary for things like the legend and the observation type on the Y-axis.

Derryrover commented 3 years ago

@lisannewapstra @lexvand

Hi Lex and Lisanne,

For your information: This ticket is not ready for testing yet, but once it is ready we need it to move as fast as we can. I will let you know

Derryrover commented 3 years ago

@lexvand @lisannewapstra

Hi Lex and Lisanne I just deployed to staging v0.3.36. Previous version on prod and staging is v0.3.35

Could you test what is possible to test on staging? Thanks in advance

kr Tom

lisannewapstra commented 3 years ago

https://nxt3.staging.lizard.net/floodsmart/Scenario1/full/23 Looks good!

Also tested other tiles and checked F12. And tested plotly functionalities.

Derryrover commented 3 years ago

deployed to production 0.3.36 graphs look good. Some graphs are still not filled, but this might be due to the missing raster.

Will contact Joeri and Jeroen

Derryrover commented 3 years ago

email to Joeri and Jeroen:

Hi Joeri en Jeroen,

Ik heb zojuist een nieuwe versie van parramatta naar productie gezet als oplossing voor dit ticket:

https://github.com/nens/parramatta-dashboard/issues/241

Veel graphs zijn nu wel zichtbaar gevuld:

https://parramatta.lizard.net/floodsmart/trainingscenario1

Het issue is hiermee echter nog niet opgelost. Zie bijvoorbeeld de empty graphs here:

https://parramatta.lizard.net/floodsmart/trainingscenario2

Dit lijkt te komen doordat dit raster mist:

b1153ca3-1268-44f2-9df1-ef0666ac4499

Naar ik begrijp zijn jullie bezig dit raster te veranderen naar een ander raster dat nog wel bestaat.

Laat me weten als ik nog iets kan doen

Derryrover commented 3 years ago

According to Joeri, Jeroen did remove the uuid of the missing raster from the client config:

Issue looks solved -> I close