The 60 seconds nginx timeout is creating some issues with actions that takes some time to complete, like creating a VM with a NFS template. After 60 seconds without a backend answer, the user is presented with a 504 Gateway Timeout error. This patch increases the timeout to 600 seconds, 10 minutes. This is enough time to process all the backend requests and ensure that a 504 error will only occur in a legitimate error in the backend. The Timeout Monitor of cherrypy is also being disabled, since we're using the nginx timeout and the session timeout instead. Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com> --- src/kimchi/server.py | 5 +++++ src/nginx.conf.in | 7 +++++++ 2 files changed, 12 insertions(+) diff --git a/src/kimchi/server.py b/src/kimchi/server.py index 0db0bb3..bca2147 100644 --- a/src/kimchi/server.py +++ b/src/kimchi/server.py @@ -74,6 +74,11 @@ class Server(object): os.makedirs(directory) self.configObj = KimchiConfig() + # We'll use the session timeout (= 10 minutes) and the + # nginx timeout (= 10 minutes). This monitor isn't involved + # in anything other than monitor the timeout of the connection, + # thus it is safe to unsubscribe. + cherrypy.engine.timeout_monitor.unsubscribe() cherrypy.tools.nocache = cherrypy.Tool('on_end_resource', set_no_cache) cherrypy.tools.kimchiauth = cherrypy.Tool('before_handler', auth.kimchiauth) diff --git a/src/nginx.conf.in b/src/nginx.conf.in index 3a8dfc5..1d1a398 100644 --- a/src/nginx.conf.in +++ b/src/nginx.conf.in @@ -38,6 +38,13 @@ http { access_log /var/log/nginx/access.log main; sendfile on; + # Timeout set to 10 minutes to avoid the 504 Gateway Timeout + # when Kimchi is processing a request. + proxy_connect_timeout 600; + proxy_send_timeout 600; + proxy_read_timeout 600; + send_timeout 600; + server { listen $proxy_ssl_port ssl; -- 1.8.3.1