Connection Occaisonally Times Out

They have: 8 posts

Joined: Sep 2005

I have a dedicated server, running...

Plesk 7.5 Reloaded
Apache 2.0.51
PHP 5.0

I wouldn't say I know a whole lot about Apache and administering a web server, but I know enough to get by.

ANYway - here's my problem. Most of the time (we'll say 98%) my web server runs fine, no problems, everyone's happy. But in that 2%, I keep getting either "Connection timed out" (Firefox) or "Page Cannot be Displayed" (IE). If you wait a few seconds (10-20) and then refresh, the page will load instantly.

I know it's not just me, because a lot of my users have been complaining about the temporary outages. And for the life of me, I can't figure out what's wrong.

I've been trying to locate a pattern in the outages, but I can't seem to find much. It happens when the server is under load, but also when it's not. It happens on all sites on the server, not just a select few. It *may* just be PHP (which is why I included my PHP version above) because all of my sites and the majority of my clients are coded in PHP. But I don't know that for sure, I'm just saying what I've observed.

Does anyone have any idea what that could be? Is there any way to fix it, or maybe some way I could tweak Apache/PHP in order to stop it from happening? Even just a shove in the right direction would be great, as I'm pulling my hair out here.

Help would be greatly appreciated! Thanks.

mairving's picture

They have: 2,256 posts

Joined: Feb 2001

It sounds more like a DNS problem. Are you managing your own DNS server? Always a good idea when the connection drops out to do a traceroute (tracert in Windows). This will give you a clue when and where the connection dies. It would also help to identify if it is an Apache issue or not.

Mark Irving
I have a mind like a steel trap; it is rusty and illegal in 47 states

They have: 8 posts

Joined: Sep 2005

Hmmm... that's a really good idea.

I'll run some tests the next time it starts happening, come back with some results.

Thanks!

They have: 8 posts

Joined: Sep 2005

Here's a couple results, a few interesting things. I have two computers here (neither are the dedicated server, that's located in a different state). Here's the traceroute from my Fedora comp:

Quote:
1 * * *
2 ge-2-3-ur02.wallingford.pa.panjde.comcast.net (68.86.218.185) 11.628 ms 14 .410 ms 11.452 ms
3 po70-ar01.wallingford.pa.panjde.comcast.net (68.86.208.190) 14.651 ms 13.5 63 ms 11.048 ms
4 po90-ar01.401nbroadst.pa.panjde.comcast.net (68.86.208.29) 12.809 ms 12.62 9 ms 12.829 ms
5 68.86.211.14 (68.86.211.14) 16.466 ms 15.877 ms 14.670 ms
6 12.118.114.9 (12.118.114.9) 13.156 ms 15.658 ms 13.935 ms
7 tbr1-p012301.phlpa.ip.att.net (12.123.137.62) 17.510 ms 16.240 ms 15.893 ms
MPLS Label=31306 CoS=0 TTL=1 S=1
8 tbr1-cl8.n54ny.ip.att.net (12.122.2.17) 17.359 ms 15.860 ms 16.403 ms
MPLS Label=32363 CoS=0 TTL=1 S=1
9 12.122.82.193 (12.122.82.193) 16.296 ms 17.290 ms 18.886 ms
10 sl-bb20-nyc-12-0.sprintlink.net (144.232.8.49) 16.697 ms 16.692 ms 17.423 ms
11 sl-bb23-nyc-8-0.sprintlink.net (144.232.7.14) 18.180 ms 18.422 ms 18.771 ms
12 sl-gw30-nyc-1-0.sprintlink.net (144.232.13.24) 15.277 ms 15.410 ms 17.611 ms
13 sl-schlu4-1-0.sprintlink.net (144.232.228.6) 50.120 ms 49.308 ms 49.558 m s
14 a0nycc2.gw-core-a.whn.schlund.net (217.160.229.106) 71.909 ms 54.507 ms 5 3.165 ms
15 * * *
16 * * *
...etc....
30 * * *

The first thing that I notice is that it never reaches its destination (which I'll show in the next one). I ran it a few times, same result. Then here's tracert from windows:

Quote:
Tracing route to developerstop.com [82.165.239.196]
over a maximum of 30 hops:

1 * * * Request timed out.
2 18 ms 11 ms 10 ms ge-2-3-ur02.wallingford.pa.panjde.comcast.net [68.86.218.185]
3 12 ms 13 ms 18 ms po70-ar01.wallingford.pa.panjde.comcast.net [68.86.208.190]
4 12 ms 12 ms 11 ms po90-ar01.401nbroadst.pa.panjde.comcast.net [68.86.208.29]
5 16 ms 15 ms 15 ms 68.86.211.14
6 13 ms 13 ms 14 ms 12.118.114.9
7 18 ms 16 ms 18 ms tbr1-p012301.phlpa.ip.att.net [12.123.137.62]
8 18 ms 16 ms 17 ms tbr1-cl8.n54ny.ip.att.net [12.122.2.17]
9 36 ms 42 ms 21 ms 12.122.82.193
10 17 ms 16 ms 17 ms sl-bb20-nyc-12-0.sprintlink.net [144.232.8.49]
11 16 ms 16 ms 16 ms sl-bb23-nyc-8-0.sprintlink.net [144.232.7.14]
12 15 ms 15 ms 16 ms sl-gw30-nyc-1-0.sprintlink.net [144.232.13.24]
13 49 ms 51 ms 50 ms sl-schlu4-1-0.sprintlink.net [144.232.228.6]
14 61 ms 55 ms 56 ms a0nycc2.gw-core-a.whn.schlund.net [217.160.229.106]
15 53 ms 53 ms 56 ms v993.gw-prtr-a.whn.schlund.net [217.160.229.38]
16 56 ms 54 ms 55 ms u15185602.onlinehome-server.com [82.165.239.196]

Trace complete.

The onlinehome-server.com is the real destination. And the first thing I notice about this is the very first line, where it times out.

What does all this mean? Any suggestions?

mairving's picture

They have: 2,256 posts

Joined: Feb 2001

The first line really means very little. It could just be a little lag time in getting out on the Internet. Could be caused by a router. I would pay more attention to the last hops since others are having the problem. That would indicate that it is somewhere near the end of the hops that the problem lies since they would all take different paths there.

Looks like this line is key:
15 53 ms 53 ms 56 ms v993.gw-prtr-a.whn.schlund.net [217.160.229.38]
since that is the line that it timed out on. That appears to be the 2nd to last hop which should be DNS. I would email whoever is doing your dedicated hosting armed with a few traceroutes and see if they can help you. You can at least prove to them a connection problem not a Apache problem.

Mark Irving
I have a mind like a steel trap; it is rusty and illegal in 47 states

James's picture

He has: 127 posts

Joined: Dec 2005

Looks like your provider might need some network attention. Ask them to look into it.

They have: 8 posts

Joined: Sep 2005

I called tech support and explained everything to them (including the traceroutes). They told me it was an Apache error, and to find the 404 errors in my access log and forward them to them so they could inspect the "failed requests".

Needless to say, all the 404 errors in my access log had to do with files not existing, nothing about connections timing out. I emailed him, but for some reason I have a strong feeling I'll be making another phone call tomorrow anyway.

/sigh

Edit: Oh, by the way, that's not my first time attempting to contact tech support about it. The first time the lady sounded like she had no idea what she was talking about, but she said she'd forward it to an admin. The guy I got today at least understood what I was trying to say, and told me that the other lady never forwarded anything. Priceless.

James's picture

He has: 127 posts

Joined: Dec 2005

Perhaps you didn't have to call them Smiling An email should have done the job with the traceroute paste. But hopefully they will get it sorted for you! If not keep phoning, support is part of your service you purchased. Best of luck.

free-space.net: free cpanel web hosting for new webmasters
sitemanaged: click here to visit our site
Premium quality web hosting for the elite webmasters and hosts. (ThePlanet and Savvis networks)
Contact me for a quote today and discuss your requirements.

Renegade's picture

He has: 3,022 posts

Joined: Oct 2002

Working in a call centre, I can tell you now that most people don't really know what's going on Sticking out tongue

Want to join the discussion? Create an account or log in if you already have one. Joining is fast, free and painless! We’ll even whisk you back here when you’ve finished.