Slow site performance?
- Thread starter snowhiker
- Start date
Stereodude
Not really a
I had a few more of the same sort of hiccups today as well. I was on a Verizon LTE mifi (again) and I found one or two other sites were equally slow, but everything else was quick.
That sounds like it might be something on Verizon's end if you're noticing other sites being equally slow? I looked through the performance logs for today and I didn't see any abnormalities. They also don't paint the entire picture so it's possible there might be network issues happening.
Stereodude
Not really a
Maybe, I only really noticed one site that was also unloadable around the same time the forum was "pausing" for several minutes when I would click on a link.
This: http://www.fox.com/gotham
I only mention that because it was the first other site I found that wouldn't load. I don't recall if there was another.
timwhit
Hairy Aussie
That Fox site is slow for me on Comcast Business internet. I think we have at least 100Mbsp downstream.
Howell
Storage? I am Storage!
That site took 10 seconds to load on my phone.
1: Since we are talking about hard to track down problems, I spent a week off and on tracking down a problem with the following symptoms: almost every night in the 3am hour connectivity over the mpls link would die for about 15 seconds. Turns out, the installation engineer configured the same stp priority on the new routers down there as our core router. Something must have been happening in isp network to cause recalculation in the 3am hour.
2: Eventually packet traces showed me a bug in our firewall when dhcp relay was enabled. When the client went to renew its address half way through the lease time, the server would see the unicast request and then the relay agent would send a request from each of its ips in the vlans, 2. So the server would send an ack, ack, nack. Also the client was an Avaya phone with a dhcp client that is not very robust the expiration of the lease would cause the phone to reboot.
The vendor is working on what they called cowboy code.
1: Since we are talking about hard to track down problems, I spent a week off and on tracking down a problem with the following symptoms: almost every night in the 3am hour connectivity over the mpls link would die for about 15 seconds. Turns out, the installation engineer configured the same stp priority on the new routers down there as our core router. Something must have been happening in isp network to cause recalculation in the 3am hour.
2: Eventually packet traces showed me a bug in our firewall when dhcp relay was enabled. When the client went to renew its address half way through the lease time, the server would see the unicast request and then the relay agent would send a request from each of its ips in the vlans, 2. So the server would send an ack, ack, nack. Also the client was an Avaya phone with a dhcp client that is not very robust the expiration of the lease would cause the phone to reboot.
The vendor is working on what they called cowboy code.
The site was down. There were some issues with the capacity configured for apache. I re-tuned it and restarted the server. The server was also at 100% swap usage and close to 100% memory usage. I think the balance between MySQL and apache was a little off.
Turns out to be a small DoS attack is hammering the server today bringing down the various services due to limited memory. I've put some changes in place to mitigate it for now but I'll need to do further investigation on this. Hopefully this stops the issue for now. I'll keep an eye on it.
Stereodude
Not really a
Ruh, roh... You've upset the North Koreans!
I may have. It wasn't a DoS against SF's domain, it was against another I host. It's originating from a single IP address at the moment which looks to be from Romania. I've since blocked it at the firewall for now to see if they come back to do more later on different IPs.
timwhit
Hairy Aussie
Works great for me.
Stereodude
Not really a
It was super slow for me earlier today when I was on a 4G hotspot. Other sites were loading fine.
ddrueding
Fixture
No issues here.
The site is still super quick and responsive.
Only problem is this.
It rare. Only once a month or so. Probably a bug somewhere in the software. I've been trained to copy my long replies to notepad so I don't lose those long replies.
Again, thanks for the site, as well as its operation, maintenance, and upkeep.
Only problem is this.
It rare. Only once a month or so. Probably a bug somewhere in the software. I've been trained to copy my long replies to notepad so I don't lose those long replies.
Again, thanks for the site, as well as its operation, maintenance, and upkeep.
timwhit
Hairy Aussie
I haven't noticed anything, but this doesn't sound good: http://tech.slashdot.org/story/16/01/01/2321211/linode-under-ddos-since-christmas
We've had some outages in the past few days because of these DDoS attacks. Our server is on the Newark node in case the site goes down and you want to get a status update on it from here. it sucks for them but I'm thinking the silver lining is that whatever group is coordinating the attack(s) is providing a free amount of distributed load that Linode can now use to harden their infrastructure.
mubs
Storage? I am Storage!
Site is reasonably fast now. This afternoon and morning (about 4 and 8 hours ago), it was unusable - didn't load at all. Last few days it's been dog slow. Just FYI.
I wasn't aware of the performance issues from the past few days but you're correct about the issues last night/early morning. I got a notification of the site being down around 2:00AM EDT. I logged in and it wasn't due to our server but rather the upstream network and routing from Linode. If you're ever interested, bookmark this page for the Linode status and look for alerts at Newark. None the less, thanks for letting me know you noticed an issue. If you're noticing performance issues, please feel free to mention it. I don't get upset by it and would rather know you're having issues so I can look into them rather than you have a frustrating experience.
I was looking at my stats and history from uptimerobot monitoring the other night and had a shake my head moment because my self-hosted website had a lot less incidents of downtime when compared to the one hosted on Linode. I realize it's not enough data to warrant that kind of change as the infrastructure I have has too many failure points...but with that said, Linode should have lots of redundancy and they still have downtime.
mubs
Storage? I am Storage!
Thanks Handy.
Slow right now.
Slow right now.
Maybe they're still working out issues. It's fast for me here right now.
My uptime monitor reported 3 hours, 16 minutes and 35 seconds. :-( That was a long outage.
mubs
Storage? I am Storage!
I think apart from your hosting provider, there are network issues. Many foreign sites were slow to dead for me yesterday - like ARS Techinca, for example. There's something screwy going on. Maybe a major DOS attack somewhere, or a network outage.
Stereodude
Not really a
About half the time when I was in China the site wouldn't even load. Other times it worked fine. I didn't bother to mention it because, well, China...
mubs
Storage? I am Storage!
Nice an snappy today!
Tea
Storage? I am Storage!
Are we running new software? On a new server? Storage Forum has always run well for me but it seems especially nippy now. Nice work Doug!
Stereodude
Not really a
But what happens when the weather is bad? Does the site go down?
But what happens when the weather is bad? Does the site go down?
It can get a bit swirly from time to time.
Stereodude
Not really a
I knew it!!! The site slowdowns are caused by thunderstorms! :bstd:
Howell
Storage? I am Storage!
This site has been very slow for the last hour. Slow like I've never seen it. Seems to have recovered now though.
Howell
Storage? I am Storage!
Ok, not completely recovered. Fast/slow minute to minute.
It timed out a few times earlier today.
Stereodude
Not really a
I use Weather Underground myself.