Production env. 504 Gateway Time-out

[Holofractographic]

NOT URGENT.
I get this message any time I click on Your projects:
504 Gateway Time-out
The server didn't respond in time.
User: Maro
It only happens on there rest of the links works. It doesnt stop me of doing things

[Xicnet]

This should be checked by @bum2.

We could increase the webserver's timeout, but it's just a temporary solution, since the problem seems to be that page loads too much data and it becomes slower as data grows.

This is a problem for scalability and should be addressed at code level, possibly implementing some caching. However, if data changes too often, caching won't be a solution since it would be required to update cache constantly and it's resources consumption would make it unsustainable.

These questions should probably help to start analysing the problem:

• Does OCP process data on a per-user basis or globally?
• Can OCP pre-process data, say every hour, so it can be cached and it should suffice?
• How long does it take to compute?
• How often could the data change?
• Would it be a problem if data changes and the users do not see the updated results until the next cache refresh?

I'm bouncing this issue to @bum2 so he can have a look.

Hope this helps!

[bum2]

The Timeout settings should be changed asap, meanwhile we improve the code. We can use the issue #509 for that, and we keep this one to talk about the changes needed in the code.

The OCP only process data for every user to render the pages she use. The timeout happens now for slow pages only when a user is coordinator of a project with hundreds of members and candidates. The coming release includes some refactoring that makes much faster the joinrequests list, but at the same time it adds some more info about the related 'exchange' (and its status) for every joinrequest with shares. Anyway i think that the new release will be faster in produccion compared to what we have now, but i will dig again in the subject to find ways to improve the speed of such big pages and reduce again the timeout settings if needed.