[09:24:57] <ema>	 looking good!
[09:24:59] <ema>	 https://grafana.wikimedia.org/dashboard/db/varnish-machine-stats?panelId=58&fullscreen&orgId=1&var-server=cp4010&var-datasource=ulsfo%20prometheus%2Fops&from=1494404528836&to=1494664671063
[09:31:41] <ema>	 usage-wise, the current maximum is ulsfo upload at ~2.5G \o/ 
[12:43:33] <bblack>	 there was a wierd spike around 04:14, probably some separate issue
[12:44:16] <bblack>	 the spike only show up well on ulsfo upload fe+be and eqiad upload be
[12:44:42] <bblack>	 it must have been traffic-induced somehow, but I donno what it was
[12:45:06] <bblack>	 (spike of transient usage up to unreasonable values and then right back down)
[12:46:29] <bblack>	 anyways, after some days of data we can set some conservative caps that leave plenty of headroom
[12:46:41] <bblack>	 maybe make them part of the FE mem calculation and get some more FE mem out of it
[12:47:44] <bblack>	 given the whole BBR comparison thing starts monday, I think we should probably leave it all alone for now
[12:48:04] <bblack>	 (ditto reprovisioning ulsfo caches, finishing maps->upload merge, etc)