Re: Status of the build.gnome.org service

[Colin Walters <walters verbum org>]

On Thu, Dec 3, 2015, at 09:55 AM, Andrea Veri wrote:

> On this side the whole service seemed to be down for around two days 
> (from the 1st to the 3rd) and the only way we noticed the outage was 
> someone joining #sysadmin and reporting the downtime.

I think what happened is `git fetch` got stuck again on
https://gitlab.com/groups/uhttpmock
I need to add timeout support.  Ideally detecting this case - I think
gitlab can sometimes start a response where the TCP connection
stays alive but the server side never sends anything.

> What we currently have handy is the following howto wiki page [1] 
> listing a series of commands to execute to seemingly start / restart 
> the service itself but we don't have any additional knowledge on how 
> to monitor the service nor any documentation to look at to properly 
> debug the build tools properly.

Yeah, right now the service runs via tmux.  Doing this properly basically
means a systemd unit file.  (Or investigating running inside a Docker
container etc.)