Observatory Rating at Instance Directory is not being updated for newly added instances

[anselmicz]

Describe the problem/question

Hi.

I have recently added my instance to the public directory at Instance Directory. However, mine and a couple of other instances do not seem to get updated with Mozilla Observatory Rating.

Is this a known issue? Is there a plan for either updating the discovery process, documenting it in the About section or documentation, or possibly dropping the Observatory Ratings in the future?

Did you use the FAQ section?

• Yes, I have read the FAQ and I found no solution/answer there.

What you did?

1. Created a new PrivateBin instance, complete with SSL certificate, DNS records, cfg/conf.php configuration etc.
2. Added instance address to Instance Directory
3. Waited 24h+ for automatic re-evaluation as stated under When do the instances get updated in the About section
4. Instance Directory still shows no Mozilla Observatory rating, even though several days have passed

What happens

Observatory Rating is not being added to multiple instances at Instance Directory.

What should happen

Observatory Rating is added to new instances after automatic re-evaluation.

Additional information

This may or may not originate from the recent change to how Mozilla Observatory works: scan results are no longer being made public. See Observatory FAQ here and here.

Server address

https://paste.trove.cz

Server OS

Debian 11

Webserver

Nginx

PrivateBin version

v1.7.6

Browser and version

No response

Local operating system and version

No response

Issue reproducibility

No, I cannot reproduce it on https://privatebin.net.

[elrido]

I have recently added my instance to the public directory at Instance Directory. However, mine and a couple of other instances do not seem to get updated with Mozilla Observatory Rating.

Is this a known issue? Is there a plan for either updating the discovery process, documenting it in the About section or documentation, or possibly dropping the Observatory Ratings in the future?

Yes, see the issue and latest update in PrivateBin/Directory#128 (comment) - basically I'm looking for any input in how I can diagnose what is going wrong.

When I run unit tests at home with a handful of instances it all works. When I manually run single API requests using wget on the VM this is hosted on, it also works. When the nightly cronjob launches the parallel async hyper client update off all instances, some (very few) do get updates, most requests simply time out or get empty responses. I suspect that the new API uses some kind of rate limiting that doesn't like our service opening many parallel connections? Any suggestions on how to diagnose this further and/or resolve this are highly appreciated. I'd hate to have to switch to polling the service in serial, like I did in the beginning - that used to take 15 minutes, instead of the 45 seconds it takes at the moment.

[anselmicz]

While I'm not a programmer by any means, my test is able to finish in under a minute and quite reliably (granted I tested it only on two geographically separate machines):

curl --parallel --parallel-max 10 --parallel-immediate --config urls.txt -X POST 1>mdn-observatory-api.log

Higher number of parallel connections starts generating 5XX error codes and timeouts.

urls.txt is a list of API requests containing all PrivateBin instances I scraped off of the Instance Directory (well, 228 of them, I haven't verified the count).

The official repository seems to be HTTP Observatory repository. Maybe creating an issue there to get some sort of guidance might get us an official reply? There already seem to be a couple of issues regarding server-side errors or timeouts when scanning using the API.

[elrido]

Thank you, I'll give diagnostics using curl on the host a try and update the issue with my findings. mdn/mdn-http-observatory#176 sounds relevant to this. If we too get error 500s, I can extend the logic to do a small pause then retry for those cases.