PDA

View Full Version : Suggestion...



rbmay
08-11-2007, 01:38 PM
I suggest putting all resource instensive sites on the same server rather than putting them on with other low resource users. It is frustrating when it takes more than 5 minutes to access your website when you have a 5 meg connection. Not very happy right now. :(

Or could it be that the servers are being packed with clients?

hpham
08-11-2007, 09:08 PM
Have you checked your connection in both directions? Try tracerouting from your computer and from your vps account. Also try pinging in both directions. If you see packet loss then that may be a sign of network problems, and not overloading problems.

The only reason I am bringing this up is because I experienced some weird slowness that I thought was being caused by some software issue. Turned out I had major packet loss going to the server, which caused slow connections to the server.

Also check the server load average. Do a 'cat /proc/loadavg' in your vps account. You will need to do this during times of high load. Then westhost and others can offer you better advice.

It would also be helpful if you posted your vps version. New customers (after May or so) are getting v3 accounts, which run on a newer kernel and sphera version, which supposedly makes things a lot faster. You can easily tell what version you have by looking at your login screen. If it says "Site Manager 2.0" then its a WH2 account. If it just says "Site Manager Login" then its a WH3 account. You could also post which server you are on (the server shows up in the login page at yourdomain.com/manager). Mine is wsl06012, and I am experiencing no overloading problems.

Harrison

corvus
08-12-2007, 11:52 AM
rbmay,

I've been with westhost for 10 months now and have never seen anything like this. Care to post the URL you are trying to access so we can see if we get the same lag?

Scott

rbmay
08-13-2007, 04:41 PM
I am on https://wsl06008.west-datacenter.net. It does not do this constantly. Only at peak times in the early afternoon somedays. I am running wordpress with wp-cache enabled. I have also compressed all my images. My url is: http://monadfm.com.

I have done a ping and traceroute and everything was good.

Has anyone noticed the downtime today? Today I did a trace and ping and I was getting 50% packet loss at the westhost data center.

I am on V3 and I am on the starter business plan.

hpham
08-13-2007, 05:50 PM
I have noticed some packet loss today. It seems as though one of Westhost's providers is having some routing difficulties. From what I can tell, it depends on which path your packets take to get to Westhost.

If you come to Westhost through XO Communications then you are completely fine with no packet loss. If you come through level3 then you may experience packet loss. This started sometime last night and has not been fixed yet.

If you are relatively sure the problem is with the server you are on then try chatting with support or submitting a support ticket. Specify the approx time this happens and hopefully they will be able to find the problem.

rbmay
08-13-2007, 06:15 PM
Yes I am connecting through level3. It has been in and out all afternoon. I am sure they are working on it. Thank you for your help.

I called the load average and this is what I received:

1.56 1.44 1.71 3/1342 27367

What do these numbers mean?

prodport
08-13-2007, 09:05 PM
I called in this afternoon when my site and several others were experiencing timeouts. I was told there was a "denial of service" attack and they hoped to have the situation resolved shortly. It was.

I have a service that sends me emails when they spot outages and I received an email the night before. When I inquired about that incident, I was told it "might be" the same issue. No email complaints from readers so I suspect the outage was short.

I don't know if this relates to your issue or not. I do hope WestHost provides some details on today's incident and how they might be addressing the problem. I think just about every web hosting company experiences these issues now and then. I knew something was up as it's the first time I've heard "high call volume" referenced on the phone system. I couldn't get to chat.

But I've always stayed with those companies that talk about the issues. I came to WestHost from a previous firm that swept these incidents under the carpet and wouldn't acknowledge them even though the outages were for hours, not 15 minutes like I've been experiencing. I think the best explanations (Anatomy of a Disaster) I've seen have been from the DreamHost people when they have had problems. Since they don't offer phone support, I don't use them for production.

I've been extremely happy with WestHot for the past two years. I recommended another company to them just last week.

hpham
08-13-2007, 11:50 PM
The /proc/loadavg numbers mean:

0.23 0.26 0.21 1/715 3007

0.23 - 1 minute average
0.26 - 5 minute average
0.21 - 15 minute average

1/715 - 1 process active out of 715 total processes - this is an instantaneous number from the time loadavg is read

3007 - last process id run (pid)

Load averages of 1 - 15 are probably ok for the vps environment. Numbers constantly in the 30s are probably a warning sign of problems, although I am not absolutely sure since I have not had lots of experience with the WH3 environment.

Harrison

corvus
08-14-2007, 04:43 PM
I am on https://wsl06008.west-datacenter.net. It does not do this constantly. Only at peak times in the early afternoon somedays. I am running wordpress with wp-cache enabled. I have also compressed all my images. My url is: http://monadfm.com.


Okay, I've got a guess here. It depends on how many people are listening in over the streaming audio stuff, but it could be possible that you are maxing out on the number of apache processes and threads.

You've got keep alives turned on in your apache conf. In your pop-up streamer, you show the album art. Now depending on the song length, you could have every user who is streaming holding open a connection to the webserver for an indefinite amount of time.

Now, more likely is that when a new song starts streaming, all the clients make there requests and then hold the connection open for 2 minutes during and this could be causing the problem. On a series of shorter songs though, they may not release the connection for a while.

Scott

rbmay
08-19-2007, 09:33 PM
My Stream was elsewhere when I was having the problem. I have it here temporarily. Are you referring to the stream or http processes?

I think it was a MySql database I was using for stats. Since I did away with that script the problem has gone. So far, so good.

Thank you,
Robin

corvus
08-20-2007, 07:26 PM
My Stream was elsewhere when I was having the problem. I have it here temporarily. Are you referring to the stream or http processes?

I think it was a MySql database I was using for stats. Since I did away with that script the problem has gone. So far, so good.

Thank you,
Robin

Good to hear that you think you've got it with the database issue.

What I was thinking of was the images of the CD's that appear when someone is streaming your music in a web browser. however it doesn't sound like this was it so...

Scott