All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Going mad with erratic website performance, help! mtr?
I have done everything possible in terms of speeding up my magento store, mysql, apache, varnish tuning, etc hosted on hosthatch uk vps . On top I even have cloudflare edge cache but frequently especially on the admin side where the cf edge cache does not work the website almost goes unresponsive, The behavior is very erratic, sometimes the load times are great but most of the time they are just downright awful, we are talking about root document loading in 30-40 seconds whereas using profiler I can see that the page is being generated in less than half a second most of the time bcoz of varnish(non admin) even without it (admin) its usually under a second. Cloudflare starts throwing 520 errors when the performance gets really awful. Wt has been confusing me is that during the said period my ssh connection usually works fine and I don't see any resource utilization at all, disk, cpu, network just seem to be idling bcoz there is practically no load but on my browser it feels like its end of the world and I am back on my 14kbps phone line modem 
I have been doing some mtr numbers which show some losses . I have tried it from Netherlands & the US. Can anyone tell me if this explains the website behavior ? But even if it does or does not plz still suggest some solutions.
From NL
Host Loss% Snt Last Avg Best Wrst StDev 50.118.225.2 0.00% 2082 0.1 0.1 0.1 53 5 146.19.169.93 3.30% 2082 1.8 5.4 0.7 90 13.4 ams-nkh-cr5.zet.net 0.00% 2082 3.5 4.6 3.4 82.6 7.9 ams-eq6.tr1.zet.net 0.00% 2082 3.6 4.7 3.5 78.1 7.8 adm-b10-link.ip.twelve99.net 0.00% 2082 3.9 5.4 3.8 101.3 8.5 adm-bb2-link.ip.twelve99.net 88.60% 2081 4 5.5 3.9 76.7 7.2 ldn-bb2-link.ip.twelve99.net 41.00% 2081 9.7 10.4 8.7 78 7.9 slou-b2-link.ip.twelve99.net 0.00% 2081 20.2 11.1 9.5 81.7 8.3 hosthatch-ic-379081.ip.twelve99-cust.net 0.00% 2081 133.3 52.6 13 428.2 41.5 my_hosthatch_vps_ip 0.00% 2081 9.4 10.5 9.2 81.8 7.6
From US
Host Loss% Snt Last Avg Best Wrst StDev irb-1110.edge1.lblkwa.crunchbits.net 0.00% 619 0.4 80.00% 0.2 25.6 2.5 edge1.spknwa.crunchbits.net 0.00% 619 0.8 70.00% 0.7 1 0.1 e0-6.core1.ska1.he.net 10.50% 619 25.8 2360.00% 1.6 27.4 3.8 100ge0-36.core1.mso1.he.net 12.80% 618 31.5 3050.00% 8.6 34.4 3.8 100ge0-36.core2.yyc1.he.net 15.70% 618 40.8 3950.00% 17.3 51 4 (waiting for reply) 48.80% 618 54 5700.00% 31.1 137.8 10.9 port-channel3.core2.sea1.he.net 48.80% 618 54 5700.00% 31.1 137.8 10.9 sea-b1-link.ip.twelve99.net 11.70% 618 43.7 41.8 19.6 62.6 3.9 chi-bb2-link.ip.twelve99.net 27.30% 618 87.8 85.2 63.4 96.3 3.8 nyk-bb2-link.ip.twelve99.net 19.40% 618 103 101.3 79.4 109 3.4 ldn-bb1-link.ip.twelve99.net 18.00% 618 175.2 173.2 152.1 178.7 3.6 slou-b2-link.ip.twelve99.net 11.80% 618 178 176.4 154 231.9 5.9 hosthatch-ic-379081.ip.twelve99-cust.net 14.90% 618 204.4 215.9 172.2 497.1 41.8 my_hosthatch_vps_ip 12.10% 618 175.5 173 150.9 176.9 3.6
Comments
To begin with, I'd suggest testing on another VPS other than Crunchbits'. Their HE link is heavily congested, which ends up with that massive packet loss.
but wt about the hostslick one in netherlands, thats the closest to the uk location. I am not very savvy with these numbers but r these numbers bad enough to impact performance of a website network wise ?
edit: I am also seeing similar loss numbers from Singapore.
I had replied to that too, but seems like my extra post content got eaten up by something. Oh well.
The NL one looks fine. A little bit of varying latency noticeable, but no real packet loss or anything really worrying that would cause the issues you're expressing. Unless you are having throttles in TCP negotiation, but that shouldn't be happening between HostHatch <-> Cloudflare's network.
I'd be betting the problem would be in the store itself; I'd check resource utilization instead and having a heavy focus on it, plus checking database query latency etc.
You may want to talk to your upstream carriers see if they are having issues on their end. It could be that the line has some saturation on the carrier side.
Thanks. I was concerned after seeing the around 90% loss on some hops. Out of curiosity wt does actually happen to website traffic when e.g. my web response packets r travelling through these hops back to my browser, if there is a 90% loss wud that mean the server will have to keep resending the tcp packets repeatedly to make up for the losses ? wud that not hamper the performance of the website? or does the routing protocol allow for some contingency?
Seconding this @raza19 -- we had a cut this morning taking a bunch of capacity out. It's been fixed for now, but if your MTR was from anytime over the last ~10 hours there was a ton of congestion on the available circuits (in particular: HE) since the cut took out their circuit to Seattle, too.
Regarding loss in the MTR you mostly only care about first and last results. It's common for routers to not respond (lower priority task than their core job: routing packets) or if pinging respond much slower than actual latency and thats pretty expected.
luv me some @crunchbits