Error 500: SQLite table missing and segfault on OpenBSD httpd + CGI server
(1) By Jony (jonypost) on 2024-03-02 17:24:13 [source]
I'm currently encountering some challenges while setting up Fossil server on my OpenBSD 7.3 system and would greatly appreciate your assistance in resolving them.
Here's a brief overview of my situation:
I've followed all the steps outlined in the documentation, excluding the HTTPS setup and certificate configuration as I'm opting for a simple HTTP setup within my local network.
Even though I followed the instructions closely, when I try to visit the local address (http://fossil.pb.local) in my web browser, I get a 500 Internal Server Error from the web server, despite all the Fossil files are in the repo directory.
Upon checking the error log, I found the following messages:
------------- 2024-03-02 16:42:11 UTC ------------
panic: Segfault during process_one_web_page
HTTP_HOST=fossil.pb.local
HTTP_USER_AGENT=Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:124.0) Gecko/20100101 Firefox/124.0
REMOTE_ADDR=192.168.1.224
REQUEST_METHOD=GET
REQUEST_URI=/
SCRIPT_NAME=/
------------- 2024-03-02 16:42:11 UTC ------------
warning: SQLITE_ERROR(1): no such table: config in "SELECT 'fossil-' || substr(value,1,16) FROM config WHERE name IN ('project-code','login-group-code') ORDER BY name /sort/"
SQL: SELECT pathname FROM sfile ORDER BY pathname COLLATE nocase;
HTTP_HOST=fossil.pb.local
HTTP_USER_AGENT=Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:124.0) Gecko/20100101 Firefox/124.0
REMOTE_ADDR=192.168.1.224
REQUEST_METHOD=GET
REQUEST_URI=/
SCRIPT_NAME=/
I'd really appreciate some help troubleshooting and fixing this issue, and I'm happy to provide more information if necessary.
(2) By Andy Bradford (andybradford) on 2024-03-03 04:34:06 in reply to 1 [link] [source]
> I'm currently encountering some challenges while setting up Fossil
> server on my OpenBSD 7.3 system and would greatly appreciate your
> assistance in resolving them.
Are you running using httpd(8) from base?
I run it and it works fine with Fossil. I have the following in
httpd.conf:
root "/website"
location "*.fss*" {
authenticate with "/auth/htpasswd"
fastcgi socket "/run/fossil.sock"
}
I have a script in /var/www/website/projects.fss that looks like:
#!/usr/local/bin/fossil
directory: /fossils
repolist
Then in /var/www/fossils I have all the repositories (e.g. *.fossil).
Because httpd(8) runs in a chroot you need to have a copy of fossil in:
/var/www/usr/local/bin/fossil
If you don't compile it statically then you'll have to figure out all
the different libraries that it needs and copy them into the chroot as
well in the appropriate location.
You also need to have a slowcgi configuration for Fossil that opens its
socket in /var/www/run/fossil.sock.
Because I'm using HTTP AUTH, the username that httpd authenticates
can be used in my Fossil repositories. I just add a username to
each repository that matches the username that I use with HTTP and
then enable REMOTE_USER support on the repository, i.e. configure
remote_user_ok in the config table of the database which can be reached
using the UI at the /setup_access URI.
I don't think I've missed anything.
Have you looked at the httpd logs to see what's going on? What about the
slowcgi logs? What about permissions on all the various components?
Andy