Since an upgrade of the Baikal app, AgenDAV (in version 1.2) became buggy with it, with the kind of issue you reported.
The latest version of AgenDAV solves it.
I just released it today so to install it you need :
I am still in the process of analysing what went wrong with the baikal update. For example I noticed that my DAVDroid client does not sync anymore…
I did not look into the CardDAV part, but according to what you said, there might be an issue too with the roundcube carddav plugin.
Thanks for investigating and giving these details. Please keep us posted.
[2016-09-03 12:11:31] myapp.CRITICAL: AgenDAV\Exception\ConnectionProblem: cURL error 35: error:0200100D:system library:fopen:Permission denied (see http://curl.haxx.se/libcurl/c/libcurl-errors.html) (uncaught exception) at /var/www/agendav/web/src/Http/Client.php line 184 {“exception”:"[object] (AgenDAV\Exception\ConnectionProblem(code: 0): cURL error 35: error:0200100D:system library:fopen:Permission denied (see http://curl.haxx.se/libcurl/c/libcurl-errors.html) at /var/www/agendav/web/src/Http/Client.php:184)"} []
[2016-09-03 12:12:42] myapp.CRITICAL: AgenDAV\Exception\ConnectionProblem: cURL error 35: error:0200100D:system library:fopen:Permission denied (see http://curl.haxx.se/libcurl/c/libcurl-errors.html) (uncaught exception) at /var/www/agendav/web/src/Http/Client.php line 184 {“exception”:"[object] (AgenDAV\Exception\ConnectionProblem(code: 0): cURL error 35: error:0200100D:system library:fopen:Permission denied (see http://curl.haxx.se/libcurl/c/libcurl-errors.html) at /var/www/agendav/web/src/Http/Client.php:184)"} []
[2016-09-03 12:12:43] myapp.CRITICAL: AgenDAV\Exception\ConnectionProblem: cURL error 35: error:0200100D:system library:fopen:Permission denied (see http://curl.haxx.se/libcurl/c/libcurl-errors.html) (uncaught exception) at /var/www/agendav/web/src/Http/Client.php line 184 {“exception”:"[object] (AgenDAV\Exception\ConnectionProblem(code: 0): cURL error 35: error:0200100D:system library:fopen:Permission denied (see http://curl.haxx.se/libcurl/c/libcurl-errors.html) at /var/www/agendav/web/src/Http/Client.php:184)"} []
Because I saw messages about permissions I took a look to /var/www/agendav and saw that the owner is root. The directory agendav is 755 but I replace the owner by www-data.
After modifying owner the message Internal server error disapeared but I now have a nice white page and only a white page. The icon “yunohost” which permits to go back to the personal panel is not present.
I’am not familiar with web apps logs but It seems that when I try to log on agendav, the http request send an url with “/webmail”. It should be “/agendav” ? no ?
One more thing … It is not directly related to our issue but I saw that the file /var/www/agendav/web/config/settings.php is world-readable. Maybe the scripts upgrade should modify perms ? Or is it a technical issue ?
Do not hesistate to ask me for more details or helps in debugging. I’am not a guru in web apps and web languages but I have few skills in system administration. At this time I do not know where to look.
I had also the same issue, following a yunohost upgrade from v2.2 to v2.4.
I have just upgraded Agendav and now it is working fine on my side.
Thank you the upgrade juju!
Thanks for the report about the file permissions, this has now been locked down properly.
Tha latest upgrade introduces a specific agendav user to run the software, and it has write access to the least possible amount of files. Also the config, which holds the mysql credentials, is not world-readable anymore.
I also confirm DAVDroid is working fine with this baikal update. My issue was due to an experimentation with nginx ssl parameters changes, seems like I used a too restricted cipher list for my Android version. The default cipher list is fine.
I took the logs during the upgrade but because there was no error I did not post it.
I made a new upgrade as you said without any success. I get a page with the message “Internal server error”. In the /var/log/nginx/mydomain-acces.log :
There is no error in /var/log/php5-fpm.log, only lines like :
[05-Sep-2016 18:44:45] NOTICE: fpm is running, pid 2636
[05-Sep-2016 18:44:45] NOTICE: ready to handle connections
[05-Sep-2016 18:44:45] NOTICE: systemd monitor interval set to 10000ms
[05-Sep-2016 18:45:54] NOTICE: Terminating ...
[05-Sep-2016 18:45:54] NOTICE: exiting, bye-bye!
[05-Sep-2016 18:45:54] NOTICE: configuration file /etc/php5/fpm/php-fpm.conf test is successful
[05-Sep-2016 18:45:55] NOTICE: fpm is running, pid 2731
[05-Sep-2016 18:45:55] NOTICE: ready to handle connections
[05-Sep-2016 18:45:55] NOTICE: systemd monitor interval set to 10000ms
By making a grep -w 500 in my nginx log files I saw that, since the upgrade, I have many occurence of that kind of http response. Not only on agendav. Maybe it is a clue. Maybe I have something specific on my system. These “error 500” are absent from my old log file (before the yunohost upgrade). For example I now have :
put agendav in the “dev” enviromnet instead of prod
used strace following the php-fpm processes (agendav pool)
I saw that when opening agendav there an error 35 from the lib curl. This error is happening when there is a problem with ssl connection.
In the strace output I saw that the php-fpm try to read the ca certificate by reading the file /etc/ssl/certs/af6cfc12.0.
This is a symbolic link to /etc/ssl/certs/ssl-cert-snakeoil.pem'. I removed that link and re-create it to point on/etc/ssl/certs/ca-yunohost_crt.pem`
Now agendav is working well, including davdroid sync.
I think what I’ve done is ugly (maybe an update will break my workaround). I do not know how cert management is working on debian.
If someone has an explanation, I’am listening.
PS : problem with my contacts in roundcube and errors 500 (probably the same root cause) are not solved
It seems that during the roundcube update there was a problem with the database.
If in the logs /var/www/roundcube/logs/carddav.warn you see messages like : BACKEND: (update_addressbook) [1054] Unknown column 'authentication_scheme' in 'field list' you can refer to the page https://github.com/blind-coder/rcmcarddav/issues/156
The idea is to drop all the tables named carddav_* from the roundcube database then logout and login into rouncube.
On my side I lost some group assignement but all my contacts are there.
Log into mysql and drop the tables in this order (there are foreign keys)
$ mysql -u root -p
mysql> use roundcube;
mysql> drop table carddav_xsubtypes;
mysql> drop table carddav_migrations;
mysql> drop table carddav_group_user;
mysql> drop table carddav_groups;
mysql> drop table carddav_contacts;
mysql> drop table carddav_addressbooks;
Now you can log out the log in. The roundcound will be a little bit slow like everyday because of the database upgrade.
This operation also solved the error 500 and white page in roundcube
Hope this work for you and do not forget to make backup before !!!
