AnimeSuki.com Forum

AnimeSuki Forum (http://forums.animesuki.com/index.php)
-   Forum & Site Feedback (http://forums.animesuki.com/forumdisplay.php?f=2)
-   -   Problem: Site very slow (http://forums.animesuki.com/showthread.php?t=99652)

Himeji 2010-12-11 12:01

Site very slow
 
Starting yesterday evening, I'm having problems with the site often (not always) being very slow and taking a long time to load pages, sometimes even timing out :twitch:
Just now, it happened again:

The connection has timed out
The server at forums.animesuki.com is taking too long to respond.
* The site could be temporarily unavailable or too busy. Try again in a few moments.

Other sites load normally. Looking at the stats on the front page, the site isn't overloaded with users either (neither yesterday nor today).

Are you having problems with the server, or working on it, or anything else that could cause those speed problems? :confused:

Kaze 2010-12-11 12:05

Sometimes

Had the same problem

Tsuyoshi 2010-12-11 12:46

Oh good, looks like I'm not the only one then. I've been having this problem since yesterday morning. My contacts pop-up window times out once ever several minutes or so. In fact, right now, I'm lucky if it reloads itself more than 3 times.

relentlessflame 2010-12-11 16:08

Just checking in for an update... is everyone still having a problem at this time, or are things back to normal? For me personally things seem to be running great, but it's possible there could be some sort of routing-related issue.

NightWish 2010-12-11 17:05

I was wondering if it might be an issue on certain routes, but I don't know enough about our current provider to know who they peer with to make a judgement. It might be helpful if anyone experiencing problems performed a trace route from their connection, to get timings and see if there is any correlation in the routes with problems.

I have had issues with the forum's responsiveness myself for a while, but put it down to local connection issues more than anything.

If there is a problem with the server, it isn't giving up much in the way of a cause. I don't have enough statistics to hand, about the server's traffic or how the incoming requests are being handled over a period of time, to make any sort of judgement. GHDpro might be able to comment further. There are "possible SYN flood" messages, but any issues resulting should be mitigated and I have a feeling they are more likely to be a symptom than an actual attack.

Himeji 2010-12-11 22:56

Problem still persists even now :(

Traceroute result:

traceroute forums.animesuki.com
traceroute to forums.animesuki.com (216.18.221.150), 30 hops max, 60 byte packets
1 <deleted> 5.108 ms 6.033 ms 7.424 ms
2 <deleted> 9.619 ms 10.611 ms *
3 * * *
4 * * *
5 * * *
6 * * *
7 webnx.ge9-27.br01.lax05.pccwbtn.net (63.218.42.194) 328.191 ms 328.343 ms 329.085 ms
8 67-220-192-101.hosted.static.webnx.com (67.220.192.101) 167.998 ms 169.159 ms 170.871 ms
9 * * *
10 * * *
11 * * *
12 * * *
13 * * *
14 forums.animesuki.com (216.18.221.150) 170.528 ms 172.008 ms 172.935 ms

The lines with the asteriks are the direct output (not modified by me).
It looks like the lag starts at line 7 with webnx.
I ran that tracerout command several times and always got pretty much the same result.

GHDpro 2010-12-12 04:06

I'm still clueless what could be causing this issue.

There was a small problem with the webserver software, which I have restarted now (both nginx & apache). NightWish: see the Cacti load average & CPU graphs (click on the graph for more historical view) to see what I mean. However, I don't think this could have been the cause as we'd all have experienced the same problem.

I also did some traceroutes from the server itself to the posting IP used by all the people who reported the problem in this thread, and the only thing I can find in common is that they all route from nyc4.he.net (New York) to lon1.he.net (London) at some point. By comparison, the traceroute for my own IP jumps from nyc4.he.net (New York) straight to ams1.he.net (Amsterdam) -- a different route, which might explain why I haven't experienced the problem at all.

However in the traceroute Himeji posted, hop 7 (lax05.pccwbtn.net) is pretty suspicious (high ping values). Also it seems to go through a different network (PCCW Global rather than Hurricane Electric).

Anyway, to get to the bottom of this problem keep those traceroutes going, please. Also post traceroutes if you had the problem (using the same PC & internet connection before), and it suddenly works fine again (so that we can compare routes and see what has changed).

I suppose we could also use a few traceroutes from members in Europe (preferably UK) who never had a problem. But we only need a limited number of those, thanks. Also please clearly state whether you ever had a problem or not, thanks.

Spoiler for My own traceroute for comparison:

Himeji 2010-12-12 14:25

Quote:

Originally Posted by GHDpro (Post 3387775)
However in the traceroute Himeji posted, hop 7 (lax05.pccwbtn.net) is pretty suspicious (high ping values). Also it seems to go through a different network (PCCW Global rather than Hurricane Electric).

Anyway, to get to the bottom of this problem keep those traceroutes going, please. Also post traceroutes if you had the problem (using the same PC & internet connection before), and it suddenly works fine again (so that we can compare routes and see what has changed).

Still the same here with that pccw server lagging a lot. Now the webnx server following it is lagging just as badly:

...
7 webnx.ge9-27.br01.lax05.pccwbtn.net (63.218.42.194) 361.647 ms 362.829 ms 363.765 ms
8 67-220-192-101.hosted.static.webnx.com (67.220.192.101) 364.730 ms 366.422 ms *
9 * * *
10 * * *
11 * * *
12 * * *
13 * * forums.animesuki.com (216.18.221.150) 174.734 ms

I also did a traceroute to www.animesuki.org, and it looks very different, without any significant lags:

traceroute www.animesuki.org
traceroute to www.animesuki.org (95.143.193.31), 30 hops max, 60 byte packets
1 <deleted> 4.899 ms 6.259 ms 8.223 ms
2 * * *
3 <deleted> 14.526 ms 16.458 ms *
4 * * *
5 * * *
6 * lond1-rc1.linx1.ip-only.net (195.66.224.142) 29.834 ms 24.188 ms
7 semal0001-rd2.ip-only.net (82.99.29.53) 52.951 ms 54.158 ms 55.597 ms
8 62.109.44.110 (62.109.44.110) 59.782 ms * *
9 * * *
10 * * *
11 * * sesto0001-rc3.ip-only.net (213.132.112.82) 64.073 ms
12 213.132.98.114 (213.132.98.114) 98.424 ms 99.803 ms 101.025 ms
13 95.143.193.31 (95.143.193.31) 103.946 ms 105.631 ms 76.965 ms

GHDpro 2010-12-12 16:33

Well the reason a traceroute to animesuki.org would look quite different is because the torrent index and the forum are located in completely different parts of the world: the forum is hosted in California, USA and the torrent index is hosted in Sweden (=closer to Europe).

Do you get the same problem when trying to access the site of our forum host, WebNX?

Edit:
Took a more drastic move. As I mentioned above there are two webservers (nginx & Apache), where nginx was handling incoming connections and Apache handled PHP (scripts). I've now taken nginx out of the equation. Meaning, everything is now being served directly by Apache.

That will increase load (a bit) and may cause some issues. But maybe it helps... if this change didn't help much/at all, please tell me so that I can put the original configuration back.

Himeji 2010-12-14 18:11

Quote:

Originally Posted by GHDpro (Post 3388464)
Edit:
Took a more drastic move. As I mentioned above there are two webservers (nginx & Apache), where nginx was handling incoming connections and Apache handled PHP (scripts). I've now taken nginx out of the equation. Meaning, everything is now being served directly by Apache.

That will increase load (a bit) and may cause some issues. But maybe it helps... if this change didn't help much/at all, please tell me so that I can put the original configuration back.

It's clearly better now, no overly long loading times or even time-outs anymore - for now. I'll post again if the lag problems return.

Traceroute:
...
7 webnx.ge9-27.br01.lax05.pccwbtn.net (63.218.42.194) 176.861 ms 178.267 ms 179.233 ms
8 67-220-192-106.hosted.static.webnx.com (67.220.192.106) 196.637 ms 197.774 ms 199.200 ms
9 forums.animesuki.com (216.18.221.150) 200.686 ms * *

Himeji 2010-12-14 22:10

What was up with the site? About one hour after I made the previous post, the forum site went down, and I couldn't access it anymore. At first I thought the time-outs were back, but after I still couldn't access the site after a few minutes, things seemed strange :twitch:
I then went surfing elsewhere and went back here checking every few minutes, but still couldn't access it and eventually gave up the fruitless checking after about an hour :heh:
Then, I came back just now, and the site seems back to normal. What was up there just then? :confused:

GHDpro 2010-12-15 03:02

Quote:

Originally Posted by Himeji (Post 3392068)
Then, I came back just now, and the site seems back to normal. What was up there just then? :confused:

Well according to both Hyperspin (server monitoring service) and Cacti (which can make remote bandwidth/CPU etc graphs through SNMP) absolutely nothing happened at all: no outage warnings from Hyperspin (which is set at 5 min intervals) and uninterrupted graphs in Cacti (that don't show any anomalies in those graphs either).

Also checked the logs, which didn't contain much issues either (other than a few issues being detected by Suhosin, a PHP security hardening plugin). Those Suhosin issues didn't involve your IP however.

Anyway... I still don't get what is going on...

Himeji 2010-12-15 17:01

Quote:

Originally Posted by GHDpro (Post 3392370)
Well according to both Hyperspin (server monitoring service) and Cacti (which can make remote bandwidth/CPU etc graphs through SNMP) absolutely nothing happened at all: no outage warnings from Hyperspin (which is set at 5 min intervals) and uninterrupted graphs in Cacti (that don't show any anomalies in those graphs either).

Also checked the logs, which didn't contain much issues either (other than a few issues being detected by Suhosin, a PHP security hardening plugin). Those Suhosin issues didn't involve your IP however.

Anyway... I still don't get what is going on...

Very strange indeed. It started suddenly, and I couldn't access the forum site at all for over an hour. Possibly it was inaccessible for longer even, as I only came back about two hours after I stopped checking.
If there are no problems with the AS server, perhaps one of the intermediaries kicked the bucket for a while... don't know what else it could have been. Other sites still worked fine during all the time.

CrowKenobi 2011-01-29 20:26

Is anyone else having the same problem I'm having? The forum is running very slow today for me (feels like I'm back on dial-up!) with pages taking forever to load (if I don't have to reload them because not all the page loaded). :(

It seemed to have started yesterday (Friday) since that's when I noticed that the forum started to load slowly (mixed in with fits of normal quick loading)...

felix 2011-01-30 05:34

Quick, run some traces to the forums.

CrowKenobi 2011-01-30 11:30

It seems to be running "normal" atm, but here's the edited trace:

7 42 ms 43 ms 44 ms 10gigabitethernet3-2.core1.den1.he.net [184.105.213.86]
8 66 ms 68 ms 72 ms 10gigabitethernet1-4.core1.lax2.he.net [72.52.92.38]
9 267 ms 200 ms 207 ms 216.218.213.250
10 67 ms 79 ms 68 ms 67-220-192-106.hosted.static.webnx.com [67.220.192.106]
11 69 ms 77 ms 76 ms forums.animesuki.com [216.18.221.150]

Why at #9 it balloons there, I have no idea...

MeoTwister5 2011-01-30 11:41

Tried a trace from over here in SE Asia.

From a command prompt traceroute, I'm getting a big spike at the 6th interval.

6 341ms 208ms 208ms webnx.com.any2ix.coresite.com [206.223.143.172]

After that it's 100+ and 200+ ms for the next routes.

milan kyuubi 2011-02-03 05:43

Just now I was unable to enter the forum, it said ''Database Error''

http://img46.imageshack.us/img46/6797/snapshoth.th.jpg

So what happened?

NightWish 2011-02-03 05:52

The forum was making too many connections to the database. Eventually it couldn't accept any more so some people will have seen that error. It was my mistake, shouldn't have happened; a query I ran, to fix three threads that you PMed me about ;), took a very long time to run. I probably missed an index, which caused the database to perform a table scan (we have over 3 million posts on this forum so they're always very slow), which backed up other requests.

milan kyuubi 2011-02-03 06:22

^Thanks for information :) Trough I was amazed that my action (if I didn't tell you about the threads then this wouldn't have happened) caused this :D

I am just that good :p :heh:


All times are GMT -5. The time now is 13:47.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.