HTTPError: 503 Server Error: Service Unavailable for url: https://coronavirus-tracker-api.herokuapp.com/v2/locations?country_code=BD&source=jhu

[mobassir94]

if i try this :
import COVID19Py
covid19 = COVID19Py.COVID19()
loc = covid19.getLocationByCountryCode("BD")

i get error : HTTPError: 503 Server Error: Service Unavailable for url: https://coronavirus-tracker-api.herokuapp.com/v2/locations?country_code=BD&source=jhu

[Kamaropoulos]

It turns out this is not an issue with COVID19Py but with the API it pulls data from.

According to ExpDev07/coronavirus-tracker-api's Issue #264, there was an outage and it seems like it's still down at the time I'm writing this.

A temporary workaround would be to use one of the live forks provided by the community on the issue's comments. In order to do that, initialize the COVID19 object as follows:

covid19 = COVID19Py.COVID19(url="<YOUR_ALTERNATIVE_URL_HERE>")

For example, in order to use @toxyl's load-balanced API, you could do the following:

covid19 = COVID19Py.COVID19(url="https://cvtapi.nl")

I hope that helps.

It might also be a good idea to keep a list of mirrors somewhere and have the package use any that is available at the time. This should sort out the issue of the default URL being unavailable and requiring the end-user to provide their own URL every time there is an outage.

[Kilo59]

@Kamaropoulos @mobassir94 it was only down for about an hour.
Even still it's probably a good idea to have a backup if you need guaranteed availability and/or cache frequently.

[proftc]

Does anyone know of any other mirrors because the jhu site is often very busy and requests seem to fail quite often

[toxyl]

Did you try my mirror at https://cvtapi.nl? there aren't many users on it, so failing requests shouldn't really be an issue.

[fffelix-jan]

I have this issue too. It also seems to be down for me.

[jdeweese1]

Hi all,

It looks like the data source is now broken.

However, on this issue on the data source repo , they recommended https://covid-tracker-us.herokuapp.com as another data source. I plugged that into the constructor in the latest.py file,

import COVID19Py

covid19 = COVID19Py.COVID19(url='https://covid-tracker-us.herokuapp.com')

print(covid19.getLatest())
print(covid19.getLocationByCountryCode(country_code='US'))

{'confirmed': 37180308, 'deaths': 1072306, 'recovered': 6091449}
[{'id': 242, 'country': 'US', 'country_code': 'US', 'country_population': 327167434, 'province': '', 'last_updated': '2020-10-12T04:54:49.884370Z', 'coordinates': {'latitude': '40.0', 'longitude': '-100.0'}, 'latest': {'confirmed': 7717932, 'deaths': 214370, 'recovered': 0}}]

It seems to pass at least the basic smoke tests. If this is indeed API compatible with our wrapper, then it may not be a bad idea to use it as our new default data source.

[Kamaropoulos]

Agreed @jdeweese1.

In the meantime, API sources have proven to be unstable so maybe it might be time to implement source mirrors support?

[jdeweese1]

I went ahead and made a PR using another data source I found #20 .

If we are able to find some more mirrors, that could definitely be a good move for us in the case that another backend fails.

[Kamaropoulos]

#20 looks good, just merged it.

https://cvtapi.nl/ also seems to be still available so maybe we could start with these two. In case something happens with the current default mirror, we'll have an alternative that will keep things running without troubling this project's users.

Then, we can look around and find what other mirrors are available, I might set up one too. We could even propose a heartbeat system for the API to have every running mirror report its status every now and then. This way, mirror discovery would be automatic.

[Kamaropoulos]

Additionally, something like this could allow us to select a mirror for the user based on their ping to it or even implement load balancing between the mirrors, even though it will probably not make any difference at this scale.

[toxyl]

Yes, https://cvtapi.nl/ is still up, but it's not load balanced anymore. I didn't have enough traffic to warrant that. Implementing a load balancing scheme between mirrors sounds like a good idea, it's cheaper for people to spin up a non-loadbalanced instance, a $5 DO droplet does the trick.

[wobsoriano]

#20 looks good, just merged it.

https://cvtapi.nl/ also seems to be still available so maybe we could start with these two. In case something happens with the current default mirror, we'll have an alternative that will keep things running without troubling this project's users.

Then, we can look around and find what other mirrors are available, I might set up one too. We could even propose a heartbeat system for the API to have every running mirror report its status every now and then. This way, mirror discovery would be automatic.

Instead of using a API, just create a github action that generates a json file based on a cases csv like this and just call that json file directly in your app. You can check my implementation here https://github.com/wobsoriano/covid3d

[Kamaropoulos]

Yes, https://cvtapi.nl/ is still up, but it's not load balanced anymore. I didn't have enough traffic to warrant that. Implementing a load balancing scheme between mirrors sounds like a good idea, it's cheaper for people to spin up a non-loadbalanced instance, a $5 DO droplet does the trick.

@toxyl Right, this way we could off-load some of the weight from the API providers as we distribute it between them, while also providing the best experience possible to everyone that needs easy access to the data.

[Kamaropoulos]

#20 looks good, just merged it.
https://cvtapi.nl/ also seems to be still available so maybe we could start with these two. In case something happens with the current default mirror, we'll have an alternative that will keep things running without troubling this project's users.
Then, we can look around and find what other mirrors are available, I might set up one too. We could even propose a heartbeat system for the API to have every running mirror report its status every now and then. This way, mirror discovery would be automatic.

Instead of using a API, just create a github action that generates a json file based on a cases csv like this and just call that json file directly in your app. You can check my implementation here https://github.com/wobsoriano/covid3d

There are multiple schools of thought on this. Your point that the data can be analyzed, kept, and accessed directly from a Git repository is a valid one and many people did exactly that at the very beginning of the pandemic. But that, for some people might seem a bit too involved and complicated, having to get the data from a repo and do all the filtering/handling of the data themselves. Also, there are people that just need the deltas since the last day, and having to get all the pandemic data for all the countries is kind of a waste if all you need is three integers and a package can already provide that.

Other people will want to download the data manually to do a deeper analysis on them. Others will want to call an API by themselves and get some specific piece of data. Lastly, others will need a simple and out of the box solution that allows them to access that same information without much effort, so they can focus on what matters the most, figure out how the spread of the pandemic evolves over time and what is the real-world effect of our decisions on the fight against COVID-19. And having to implement the same thing over and over again isn't the best use of developers' and data scientists' time.

I believe that all these solutions can co-exist and be there for when people need them.

[Kamaropoulos]

#20 looks good, just merged it.

https://cvtapi.nl/ also seems to be still available so maybe we could start with these two. In case something happens with the current default mirror, we'll have an alternative that will keep things running without troubling this project's users.

Then, we can look around and find what other mirrors are available, I might set up one too. We could even propose a heartbeat system for the API to have every running mirror report its status every now and then. This way, mirror discovery would be automatic.

I also set up a mirror a couple of days ago. It's available at https://covid19-api.kamaropoulos.com