Construct proper JSON for `info` route for databases generated at IRIS

[CTrabant]

Some new model databases generated at IRIS by @sstaehler cause the info route in Instaseis to produce malformed JSON.

Here is the info-JSON for one of the new models: http://service.iris.edu/irisws/syngine/1/info?model=prem_ani_10s

The malformation is the arrays of NaN values, they either need to be quoted as strings or actually be numbers. Obviously they should be numbers.

The same Instaseis software works just fine with other models, e.g.: http://service.iris.edu/irisws/syngine/1/info?model=test

[krischer]

It's not related to Instaseis as it relies on what AxiSEM writes to the netcdf4 files.

@sstaehler @martinvandriel Should I leave this issue open or do you track it on AxiSEM's side?

[CTrabant]

The http://service.iris.edu/irisws/syngine/1/models endpoint will continue to fail until proper JSON is returned by the Instaseis servers.

I presume that Instaseis is creating JSON from values in the netcdf files. If true, Instaseis is responsible for creating invalid JSON. One suggestion to mitigate is to detect if the values are numbers versus non-numbers. If they are non-numbers create proper JSON by quoting the values.

Of course it'd be nice to have correct values in those arrays eventually...

[krischer]

I mean I could catch this is Instaseis but this is actually vital information and should be correct in the netcdf file.

@sstaehler @martinvandriel I guess you still have the simulation settings so it should be possible to fix the already existing netcdf files with almost no effort.

[martinvandriel]

Independent of instaseis creating wrong Json from valid netcdf, there seems to be something wrong with these runs: these values should not be NaNs but propper numbers and these numbers are actually needed in instaseis. We had some issues with the latest netcdf library, could this be the case here? @sstaehler

[sstaehler]

Was a problem in the AxiSEM script that transformed the stride of the large variables in the NetCDF file. I will manually correct all files relevant for the Syngine database.

[martinvandriel]

it seems the -g and -G options of nccopy interact in a way such that the order of the option makes a difference. this caused the variables to be created, but no data copied. Working on it:

https://github.com/geodynamics/axisem/commit/11ec15f7156f002213c1c6c4167c0083bfbcaa40

As the surface group should be removed, we need to move these variables to a different group. I guess we fix the files manually for now to be ready for AGU.

[sstaehler]

Agreed, on it

[sstaehler]

@CTrabant: I fixed all databases in /products/axisem/MODELS/ on dpstage

Can you please check, whether they work

[CTrabant]

Three of them still not working.

All working after Instaseis update.

model	status	URL
ak135f_5s	working:	http://service.iris.edu/irisws/syngine/1/info?model=ak135f_5s
iasp91_2s	working	http://service.iris.edu/irisws/syngine/1/info?model=iasp91_2s
prem_a_10s	working	http://service.iris.edu/irisws/syngine/1/info?model=prem_a_10s
prem_a_20s	working	http://service.iris.edu/irisws/syngine/1/info?model=prem_a_20s
prem_a_5s	working	http://service.iris.edu/irisws/syngine/1/info?model=prem_a_5s
test	working	http://service.iris.edu/irisws/syngine/1/info?model=test

[krischer]

Did you update instaseis?

On 8. Dezember 2015 17:26:16 MEZ, CTrabant notifications@github.com wrote:

Three of them still not working:

model	status	URL
ak135f_5s	NOT working:
http://service.iris.edu/irisws/syngine/1/info?model=ak135f_5s
iasp91_2s	working
http://service.iris.edu/irisws/syngine/1/info?model=iasp91_2s
prem_a_10s	NOT working
http://service.iris.edu/irisws/syngine/1/info?model=prem_a_10s
prem_a_20s	working
http://service.iris.edu/irisws/syngine/1/info?model=prem_a_20s
prem_a_5s	NOT working
http://service.iris.edu/irisws/syngine/1/info?model=prem_a_5s
test	working
http://service.iris.edu/irisws/syngine/1/info?model=test

---

Reply to this email directly or view it on GitHub: https://github.com/krischer/instaseis/issues/27#issuecomment-162935543

[CTrabant]

I restarted Instaseis, but did not update it. I presume that means there was a change? I'll update it when I get to the office.

[krischer]

Yes. The DB changes also required a small instaseis change.

On 8. Dezember 2015 17:35:48 MEZ, CTrabant notifications@github.com wrote:

I restarted Instaseis, but did not update it. I presume that means there was a change? I'll update it when I get to the office.

---

Reply to this email directly or view it on GitHub: https://github.com/krischer/instaseis/issues/27#issuecomment-162938399

[CTrabant]

Working.

Now the models endpoint works as an overview summary of Syngine models: http://service.iris.edu/irisws/syngine/1/models