Posted by alexus on June 19, 2013 at 3:15pm
we're using pressflow-6.28 on redhat6.4 w/ php5.3.3, there are two web heads residing behind varnish-3.0.4, web heads doesn't seem like experiencing heavy loads (0.2 15 min).
according to newrelic's dashboard /page_manager_node_view is:
- most time consuming (97.1%)
- slowest average response time (2050ms)
- apdex most dissatisfying (95.6%)
- highest throughput (14.4rpm)
I have apc and memcache enabled, yet to crawl about 8k pages and that's after some optimization (enabling cache on some of views) I was able to get down to 237min instead of 460min (before enabling cache on views).
.. this is too much .. any suggestions?
Comments
newrelic is telling you that
newrelic is telling you that you've got some slow panel pages. The low loads numbers means you're not killing your servers, but that doesn't mean that the panel pages aren't slow.
2 seconds isn't terrible, and if varnish is effectively working then it'll mainly bother just your authenticated users, but I'd guess your pages aren't very efficiently built, and the views caching experience suggests more of that kind of thing might work.
You'll probably want to try out a code profiler and/or the dev module query log to see if they pinpoint any obvious targets for improvement.
Check it at
Check it at http://www.isvarnishworking.com/ for test your varnish configuration.
while(alive){learn();}