mabass

How can I sign up for a VPS with a non-american address? Ordering with paypal or credit card fails otherwise.

That's fantastic news. Are you set financially? I'd happily contribute to a donation drive. If only to see it go beyond 17.000$ as final salute to cpanel.

Ah perfect, I wasn't sure if that would only disable my personal view or the entire logging process. Thanks.

Hi, as per title: Is this possible? Rather than dealing with DSGVO-Hassle (which would be required, as IPs are considered personal data) I'd rather just disable it and be done with it.

Thanks I'll def. look into that. For now this topic can be closed I think.

Thanks. Did you see my edit regarding profiling?

No sorry, I was unclear. I added a new subdomain audax-abc.mabass.heliohost.org, with /audax-abc/ as root. Basically the same app is now running twice, once via the sandbox (accessed by mabass.tk) and as standalone (audax-abc). I just wanted to use the standalone the same httpd conf method.

Happy to help, DENIC stopped complaining as well. My issue's also resolved, the routing now works as intended. Topic can be closed.

True. Very odd. Could you do me a favor and do the same for public_html/audax-abc ? I mirrored the specific app in it's own folder. That way I can provide a stable version while still testing. Anyway, thanks for all your help. For now I've run out of ideas, maybe stackoverflow has an idea. Edit: One thing I have not tried is profiling the app using werkzeug. However, do that I need to wrap the wsgi app and I'm not sure how to do that with your infrastructure. A basic example to deploy the profiler is provided here (under section "Profiling for Perfomance") - however the example code uses now deprecated imports.

Doing more testing, and it def. seems to be tied to server load. Right now Tommy's at 14.14 load (Edit: now at 68) and the form's break reliably, with response times around 1s. I tried pre-loading some image files to cache them user-side but that doesn't seem to help, the response delay is still too high. Have you tried switchting to the httpd.conf method? I don't know enough about Apache/wsgi to tell if this could result in a speed gain. Edit: More testing. I'm observing a correlation with htaccess and related domains. If I reduce my ht-access to the bare minium, the waiting delays seem to improve.

Thanks, though I think I fixed that specific issue shortly after. I'll have a closer look tomorrow. Anyway, thanks for taking your time. This is amazing support.

UDP Port 53 according to their docs.

Hm, I still get the issue occassionally; in any case I had switched %1 to %{Request... some time after I had first noticed the issue. It's more pronounced on the other form - any "first" time the form is submitted, the delay hits ~1s; the second submit it works fine and the download link is displayed. My best guess is that somehow my app causes a longer waiting-delay on heliohost than it does on e.g. pythonanywhere and flask's routing breaks as a result. Maybe it's related to accessing (and in the case of the form linked above: writing) files? That said, the issue seemed to be worse a few hours ago when Tommy's load was around ~10; now it's around ~5. That's anecdotal however.

Did a bit more testing with networking tools and the problem more often than not corresponds with longish server delays. The initial (failing) POST-request has nearly one full second waiting delay. After running the app successfully once, the waiting's down to ~150-200ms. Whenever, for whatever reason, the delay jumps above 500ms the app seems to break as described. Not sure if there's much to be done about it besides switching to a VPS?

Alright thanks. It's indeed a recent change. I was confused as it was properly refering to heliohost but ending up in the root public_html. Should I be concerned that DENIC is complaining about being unable to access a port?