Concerning "server level errors": With the current version of OpenACS the number of erros is very little. We have now weeks on openacs.org without a single error or with very little errors in the error.log (in the following graphic, these is the orange line)
Up to january this year, openacs.org had on average 1 error per second.
Concerning the request of Iuri: i did some tests and fixed a few issues (mostly IPv6) and the following testing scenarios work fine:
Testing setup (A):
- http-port: 8009
- https-port: 8443
- base-hostname: localhost
- alternate-hostname: hugo
- main-subsite: /
- alternate-subsite: /hugo
- host-node map: hostname "hugo" -> /hugo/
- apm-parameter (acs-tcl): UseHostnameDomainforReg: 0
In this setup, the subsite (here: /hugo/) has to be given explicit read permission for public, otherwise, one cannot login, since subsite/register won't work (one needs already permissions to reach the subsite main node). Therefore, the site-admin has to grant read permissions to the public explicitly.
Other than this, everything looks ok with setup (A), the subsite works with http and https.
Testing setup (B):
- same as above, but:
- apm-parameter (acs-tcl): UseHostnameDomainforReg: 1
When UseHostnameDomainforReg is activated:
- login is redirected to the main site; the return_url as well.
- login cookies are set on the main-site and are not available on the subsite via the browser (maybe when setting "CookieDomain" and use "domain cookies" and appropriate sub-domain host names; not tested).
- This means that out of the box the user does not stay on the subsite but has to do all authorized work on the main site (which makes management of different users on subsite useless).
Also setup (B) works as well with the newest version in the oacs-5-9 branch. There were some issues in earlier version, but the actual version is fine in this respect. No nginx is involved.
For more details. see: https://openacs.org/bugtracker/openacs/patch?patch_number=848
all the best
-g