You are browsing a read-only backup copy of Wikitech. The live site can be found at wikitech.wikimedia.org

LVS and Varnish: Difference between revisions

From Wikitech-static
Jump to navigation Jump to search
imported>Krinkle
No edit summary
imported>Ema
Line 7: Line 7:
[[esams]] is representative of all cache-only sites; [[ulsfo]] is similar.
[[esams]] is representative of all cache-only sites; [[ulsfo]] is similar.


"[[upload.wikimedia.org|upload]]" and "mobile" traffic operate similarly to "text" shown here.
"[[upload.wikimedia.org|upload]]" and "misc" traffic operate similarly to "text" shown here.


When [[LVS]] balances traffic to the :80 [[varnish]]es directly for HTTP, it choses randomly.  For :443 to the nginxes, it uses a hash of the client IP to help with SSL session persistence.
When [[LVS]] balances traffic to ports :80 ([[varnish]]es), and :443 (nginxes), it uses a hash of the client IP to help with TCP Fast Open and SSL session persistence respectively.


Within the caching layer (cp[1234]xxx machines), the jump from [[nginx]] to varnish:80 is direct on the local host only.
Within the caching layer (cp[1234]xxx machines), the jump from [[nginx]] to varnish:80 is direct on the local host only.

Revision as of 14:11, 28 November 2017

LVS and Varnish infrastructure

Wikipedia request flow

Diagram of "text" traffic flow through Wikimedia front edge LVS/nginx/Varnish infrastructure.

Notes

esams is representative of all cache-only sites; ulsfo is similar.

"upload" and "misc" traffic operate similarly to "text" shown here.

When LVS balances traffic to ports :80 (varnishes), and :443 (nginxes), it uses a hash of the client IP to help with TCP Fast Open and SSL session persistence respectively.

Within the caching layer (cp[1234]xxx machines), the jump from nginx to varnish:80 is direct on the local host only.

However, the jump from varnish:80 (frontend) to varnish:3128 (backend) is different: for that jump, we hash on the URL (etc) when balancing to the backends to divide the cache space among all machines, and thus the request typically moves from one machine to another within the cache layer here.