cache warming auth cache

Please take a look at the Cache Warmer drush extension. It supports crawling with authenticated users.

The most prominent problem I see at the moment is that authcache will not save a page to the cache if the has_js cookie is not on the request. However it is possible to simply pretend that the cookie is there by directly manipulating the $_COOKIE superglobal, e.g. in settings.php (see also #2092185: no_cache_reason: "Caching disabled for session (nocache cookie set)"):

if ($_SERVER['REMOTE_ADDR'] == "your.crawler.ip.here") {
  $_COOKIE['has_js'] = '1';
}

With that has_js tip, I am successfully using HTTPRL Spider now to crawl as a masqueraded user. It is not necessarily secure (currently relying only on ip_address() matching 127.0.0.1), but it works.

https://drupal.org/node/2172933

HTTPRL Spider in combination with XMLRPC Page Load allows you to crawl all entities on a site with any user ID of your choosing. You can also define non-entity paths to crawl, as discussed here: #2684849: Crawl Pages with specific Views Exposed Filters active

I also added a feature request to potentially allow dynamic caching based on site usage here: #2684945: Provide Cache Miss View (Uncached Pages)

Setting to Active, since no patch is attached.

I see 'XMLRPC Page Load' can not warm cache as 'xmlrpc' uses POST only method and authcache exludes such method from caching.

See: xmlrpc() -> _xmlrpc()

  $options['method'] = 'POST';

See: authcache_authcache_request_exclude()
See: authcache_builtin_cacheinc_retrieve_cache_page()

  // Only GET and HEAD requests allowed.
  if (!($_SERVER['REQUEST_METHOD'] === 'GET' || $_SERVER['REQUEST_METHOD'] === 'HEAD')) {
    return FALSE;
  }

This module Force JS does what #2 suggests and works for me when using Cache Warmer.

One caveat with using the Cache Warmer module (and any other drush-based cache warming systems) is that Authcache skips active cache backends for drush commands. When using authcache_debug, after I run the cache-warmer drush command I get in the authcache_debug dblog: "Excluded: No active cache backend."

Upon looking for the above error in Authcache, I see that the reason there is no active cache backend is due to the module ignoring drush entirely:

/**
 * Initialize the cache backend module.
 */
function authcache_backend_init($module, $vary_header, $initial_key) {
  if (drupal_is_cli()) {
    return FALSE;
  }
  ...
}

It makes sense that Authcache is not initialized during normal drush commands, but is is possible to implement a bypass specifically for cache warming drush commands?

No activity, closing this.