LVE Issues

We have been having some load issues on our server running CloudLinux 6.6 and WHM 11.42.1 b31. Every once in a while the server load would go though the roof, but without any cause like a traffic spike or runaway script that we could find. In these cases the whole server would go down. It was recommended we do a kernel upgrade so we did. It seemed to help a bit. We still have load spikes, but not to the previous degree yet. It seems like now LVE is sort of doing it's job where, when 1 site has increased traffic it gets slow (very slow) but others don't. We had LVE settings at the default. But a few things about LVE seem weird. 1) When there are faults they don't show up in the statistics page regardless of how we set the filters. (we have also turned on email notifications for faults but never get any of those) 2) When an account nears a limit (or maybe hits it) it brings down the site for a LONG time. We monitored a site in real time yesterday. It had 12/20 entry processes but new requests would take several minutes to load (even thought PHP timeout is 30 seconds) We checked the apache process logs and the same processes stayed active the whole time. During this time the LVE CPU was set to 100% limit. The site was consistently between 50% and 80% with a peak at 113%. The site in question is a pretty basic static site (although it is database driven). All the pages accessed at this time were read-only pages. Based on what I understand of LVE, requests should never get into apache if LVE feels they shouldn't. So I'm not sure why we weren't getting 508's for our requests. They were acting like they were loading for the longest time. Also I'm not sure how the requests were sitting in apache land for more than the PHP timeout length. Also not sure why the site was coming down when none of the limits were being hit that we saw. I turned off LVE and everything seems fine now. I THINK we were running LVE .8. Just wondering if anyone else has had similar issues or if an upgrade to LVE would help.