Leonard Ehrenfried's questions

I'm running a map tile server behind an nginx proxy. The tile calculation is expensive and that's why I cache the response for 24 hours. I've configured nginx to serve a stale response even when the cache is expired and at the same time updating the cache in the background.

My config is this:

  location ~* ^/tiles/(streets|satellite-overlay)/(.*)$ {
    proxy_pass          http://127.0.0.1:5000/styles/$1/$2;

    # configure server-side cache
    proxy_cache         tiles;
    proxy_cache_valid   200 1d;
    proxy_buffering     on;

    # whilst tileserver recalculates the new tile, serve the old tile instead
    proxy_cache_background_update on;
    # only one request per tile
    proxy_cache_lock              on;
    proxy_cache_use_stale         error timeout invalid_header updating http_500 http_502 http_503 http_504;

    add_header          X-Cached $upstream_cache_status;
    # cache client-side for 24 hours
    add_header          "Cache-Control" "public, max-age=86400";
  }

So a request to /tiles/streets/1/2/3.png is being correctly proxied to http://127.0.0.1:5000/styles/streets/1/2/3.png.

However, after the 24 cache period, when the cache is supposed to be updated in the background, the following URL is requested from the tile server: http://127.0.0.1:5000/styles/streets//. This leads to a 404 and the cache is never updated.

It appears that the contents of the $1 and $2 variables is being lost somewhere.

Can someone spot a configuration error? Is it possible to use these variables in proxy_pass in conjuction with proxy_cache_background_update?

I would like to use the variable $PWD in a systemd service file but it appears that it is always empty even though I have set an explicit WorkingDirectory.

[Unit]
Description=data-builder
Requires=docker.service
After=docker.service


[Service]
Restart=always
# docker-pull could take a while so we disable the timeout
TimeoutStartSec=0

WorkingDirectory=/var/data-builder/


ExecStart=docker run -e HOST_DATA=${PWD}/data -v -v ${PWD}/data:/opt/data-builder/data someorg/data-builder:latest


[Install]
WantedBy=multi-user.target

Is $PWD defined in a systemd unit? Can I use another variable instead?

I've installed Wordpress with Nginx and PHP-FPM on my own server at https://saskia.photo

The installation works great but the XML-RPC (used by Jetpack and the phone apps) throws the following error:

<?xml version="1.0" encoding="UTF-8"?>
<methodResponse>
  <fault>
    <value>
      <struct>
        <member>
          <name>faultCode</name>
          <value><int>-32700</int></value>
        </member>
        <member>
          <name>faultString</name>
          <value><string>parse error. not well formed</string></value>
        </member>
      </struct>
    </value>
  </fault>
</methodResponse>

You can see the error yourself by running

curl -A "Jetpack by WordPress.com" -is -H 'Content-Type: text/xml' --data '<?xml version="1.0"?><methodCall><methodName>demo.sayHello</methodName><params></params></methodCall>' 'https://saskia.photo/xmlrpc.php' && echo

I've worked through Jetpack's troubleshooting page but no luck.

The XML error returned leads me to speculate that the client request XML is being corrupted somewhere between nginx and PHP so that it cannot be parsed.

Here is my nginx config file

server {
  listen      80;
  listen      [::]:80;
  server_name saskia.photo;
  rewrite     ^   https://$server_name$request_uri? permanent;
}

server {
  listen                443 ssl http2;
  listen                [::]:443 ssl http2;
  server_name           saskia.photo;
  ssl_certificate       /etc/letsencrypt/live/saskia.photo/fullchain.pem;
  ssl_certificate_key   /etc/letsencrypt/live/saskia.photo/privkey.pem;

  root                  /srv/wordpress/;
  charset               utf-8;

  client_max_body_size 64M;

  # Deny access to any files with a .php extension in the uploads directory
  location ~* /(?:uploads|files)/.*\.php$ {
    deny all;
  }

  location / {
    index index.php index.html index.htm;
    try_files $uri $uri/ /index.php?$args;
  }

  location ~* \.(gif|jpg|jpeg|png|css|js)$ {
    expires max;
  }

  location ~ \.php$ {
    try_files $uri =404;
    fastcgi_split_path_info ^(.+\.php)(/.+)$;
    fastcgi_index index.php;
    fastcgi_pass  unix:/var/run/php/wordpress.sock;
    fastcgi_param   SCRIPT_FILENAME
      $document_root$fastcgi_script_name;
    include       fastcgi_params;
  }
}

Has anybody seen a problem like this or can spot a problem in my config files?

I'm using a VPS on Ubuntu 10.10 to serve my private website and Apache 2.2.16 keeps crashing without a trace. The server is on a very light load with about 300 visits/day.

To solve this I have set up monit to regularly check on apache the but that keeps crashing too.

I have checked my resource consumption and I'm well below the critical level:

I have checked /etc/apache2/errors.log and /var/log/syslog but couldn't find a trace of an error. I'm pretty much at the end of my wit.

How can I debug this issue? I'm using a pretty cheap VPS. Could they be the reason for the flakyness?

I want to convince my boss that we should be using git for version control. He says, that it absolutely must authenticate users through our central LDAP server.

I looked at the various solutions (gitweb, gitorious ... ) and couldn't really find a definitive answer about whether they support LDAP authentication.

The only solution I could find a little info on was a Apache+mod_ldap setting. But that would mean that the user authenticating on LDAP wouldn't necessarily be the same as the actual git user, right? (Not that this is a huge problem, but just something which would bug me.)

So, what's the best way to authenticate git users via LDAP?

I have a shared hosting account with SSH access. I have set up a virtual Python installation and have copies of the system folders in $HOME: ~/bin, ~/lib ...

I have installed mercurial through easy_install and everything is groovy: The hg executable sits at ~/bin/hg and my $PATH knows of this. I can create and commit to repositories fine on the server.

However, when I try to clone my repo to my laptop via SSH I get an error that hg can't be found.

$ hg clone ssh://myuser@server/hg/foobar
remote: jailshell: hg: command not found
abort: no suitable response from remote hg!

How to I tell mercurial (or SSH or jailshell) where to look for the executable?