couldn't parse data while fetching organizational rules

[tomholub]

I have an email on nvimp.com domain. With this, the app will attempt fetching rules from https://fes.nvimp.com/api/ and fail, because there is no FES running there.

Most customers will not run our server software on their domain, therefore this is an expected / normal scenario and the app should continue working.

Follow this logic when calling FES first time to determine if it's running or not in getActiveFesUrl:

call `https://fes.<domain>/api/`, process the response:
  -> did I receive a response and it's status 200, it's a JSON and has the right service value? -> use FES
  -> did I receive a response and it's status 404? -> don't use FES  
  -> did I receive a response and it's other status 4xx or 5xx? -> show error to user
  -> did I NOT receive a response (server down or timeout) and the internet is otherwise working? -> don't use FES
  -> did I NOT receive a response (server down or timeout) and the internet is not working? -> show error to user

That means if I call https://fes.nvimp.com/api/ and the server is down, but internet is working, then this user does not use FES.

[ekievsky]

#409 (review)

that PR tolerates only 3 cases:

• 404 error - did I receive a response and it's status 404? -> don't use FES
• -10001 - timeout
• 503 - server is down
  So that's the cases from your description. Did I get you wrong?

[tomholub]

again, you have to differentiate between calling FES and EKM. this issue is clearly defined as tolerating certain error scenarios when calling FES. nothing here says we should tolerate errors in calls to EKM.

…

On Sunday, July 25, 2021, Evgenii Kievsky ***@***.***> wrote: #409 (review) <#409 (review)> that PR tolerates only 3 cases: - 404 error - did I receive a response and it's status 404? -> don't use FES - -10001 - timeout - 503 - server is down So that's the cases from your description. Did I got you wrong? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#408 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABQDZEPY2JVLXABNPRIPKZDTZO6LJANCNFSM5A32X55Q> .

-- -- Tom James Holub <http://holub.me/>

[ekievsky]

ah, I messed up with abbreviations

[ekievsky]

Now it's implemented as there can't be any user without OrgRules, so what does don't use FES means, should the user proceed with empty ClientConfiguration and then try to update in on the next run of application?

[tomholub]

Now it's implemented as there can't be any user without OrgRules, so what does don't use FES means, should the user proceed with empty ClientConfiguration and then try to update in on the next run of application?

Exactly. If there is no FES running at that user's domain, then that means empty OrgRules.

To clarify futher:

call https://fes.<domain>/api/, process the response:
-> did I receive a response and it's status 200, it's a JSON and has the right service value? -> use FES to grab OrgRules
-> did I receive a response and it's status 404? -> don't use FES -> empty OrgRules
-> did I receive a response and it's other status 4xx or 5xx? -> show error to user
-> did I NOT receive a response (server down or timeout) and the internet is otherwise working? -> don't use FES, empty OrgRules
-> did I NOT receive a response (server down or timeout) and the internet is not working? -> show error to user

If the call to FES has failed, you can evaluate if internet is working or not by making one additional call to a Gmail API, for example sendAs. If you get back a success, then internet is working, and you can act accordingly (empty OrgRules). If you get back an error, then internet is down and you should show user an error describing that.

[tomholub]

For completeness, here are references to documentation: deployment overview for enterprise users who are using FES, EKM, WKD: https://flowcrypt.com/docs/technical/enterprise/email-deployment-overview.html

I'll also arrange more docs and post them here. The more you know and understand how all of this fits together, the more valuable your individual contributions to the codebase will be.