The hj website keeps \"hanging\" for up to several minutes at a time. I.e. click on a link - nothing happens whatsoever until about two minutes afterwards. It is like this for a short period and then is fine once again for ten minutes or so. Then back to hanging. Note this is the whole site, not limited to the backroom.
Anyone in charge aware of the problem or what is causing it?
Not my PC/connection - everything else fine...
|
|
Yes I get the same Dan. Yesterday it took 5 minutes to get on to discussion. At the moment very quick.
|
I've been trying to get on all day - roughly every hour or less, and this is the first time I've actually got as far as one of the forums....
I have to grow old - but I don't have to grow up
|
Its the site and I am getting quite fed up with it myself.
All I can do is ask you to bear with it. As soon as we can find what is causing it, we will fix it. It is very frustrating that after so long of no performance issues at all, we're now getting buried with them.
People are working on it. As soon as I have something useful to say, I'll say it; and as soon as we can make the problem go away, then we will do so.
Sincere apologies and we're doing our best (well, to be honest Stephen is doing his best, not much the rest of us can do).
Again, please bear with us.
Mark.
|
Well everytime I hear of issues I call up the ISP who provide the connection and they test the site out and find that it is very responsive. They are of course on the same network as the server. So if it really is a network problem with on eof the routers on the way, could I ask if you help us please?
Next time it is really slow for you :
- launch Command prompt or MSDOS prompt.
- key in tracert www.honestjohn.co.uk
- press return
If you are using Linux or a mac - perhaps you could copy and paste your results here so we can see if there is a hold up. Basically anything above 100ms response - say even up to 1000ms is a bit slow. Could you let us know what routers those are?
Thanks for your help
----------------------------------
Stephen Khoo
www.khoosys.net
|
The slowness seems to have started once the links to the oz ad thing appeared and also with the bits and pieces below the menu bar. I have done a tracert during a slow time and it comes out fine and whizzes straight to the honestjohn server. It seems to be the response of the webserver process itself which is slow. If you telnet to port 80 while it is slow then you get no response, if you do it while everything is ok you get a pretty fast response.
I think one of those http snoopers might hold the key as I think seeing what the webserver is sending back during the slow times would help.
You are not reaching a session limit/ process limit or other limit during the slow times are you?
teabelly
|
|
|
Stephen
Running really slow for me at 20:15 on 22/7.
Have run tracert on HJ no router link worse than 156ms, pinged site, average 144ms.
Compared to figures from my cricket club site which is running OK, tracert no worse than 152ms, pinging at average 143ms.
Would say from that that although site still displaying slow compared to others network appears to look good, or at least no worse than a couple of other sites which are loading at good speed. Hope this helps narrow it down a little.
Cockle
|
Stephen
Mystery deepens, site now running as quick as ever for me but tracert gives one router link as high as 277ms, linx1.lon1.as8553.net, and site pings at 158ms average.
Think you're going to need a bit of luck with this one!!
Cockle
|
What does all this mean ?
|
Maybe someone has been attempting a small scale DoS (denial of service) attack.
Very useful website on issues such as these grc.com/intro.htm
Jonathan
|
|
May be it's been serviced by a VW dealer?
|
|
|
The only DoS attack that would be likely would be the one involving cisco routers running ios 12.x I think that the hosting company would notice if their routers were locking up as a result.
teabelly
|
|
|
|
What does all this mean ?
It means that one of our hampsters has died and until we find a replacement, Mark and I are having to pedal a little bit faster to power the generator.
|
|
Can someone please explain how do I launch Command and do a tracert search? Sorry, but this is all Japanese to me.
|
|
The whole site wouldn't load for me (not just the forums) just before 7 pm tonight. VisualRoute tracing showed no untoward delays in the route, and I stick by my earlier theory about the issue being within the server rather than the comms. I don't know much about web hosting though so I could be talking out of my back end...presumably SQL or similar is involved somewhere, it's indexes in that kind of thing which I think might have gone awry. Or, as has been suggested, an internal limit being reached.
|
Thanks for your help in trace information. It looks like the network is OK.
I am sure what is happening is that the linked popop ads are slowing people down. This is because lots of you report slowdowns just after we put the link ads on.
This leaves more people in queues on the server
This causes the server to hit its limit wrt database connections (200)
and it means that the server starts getting low on memory as loads of people are in the queues.
This makes the server itself go slowly as it starts to use its virtual memory - i.e start swapping to disk. This compounds the problem.
Solutions
I asked to remove the last ad, but we are contracted to have it there till the end of the month. I have only reduced the number of pages it is on.
Even with faster hardware, we will get the same problem as we will end up with the same limits getting reached.
More memory will help - and we can also increase the internal limits. Clearly there is no point increasing those limits when the server is reaching memory limits already.
We have problems increasing the memory on the current server as most slots are hidden under the drives etc. and this would cause a long downtime.
We can only try to bring forward the replacement machines. This is not easy as all software has to be compatible with the new releases of Linux, Apache 2 etc. Some of the image libraries we use are proving a problem with this.
We are working on it though - so thanks for your patience.
----------------------------------
Stephen Khoo
www.khoosys.net
|
Can someone please explain how do I launch Command and do a tracert search? Sorry, but this is all Japanese to me.
In most Windows setups you go to Start - Accessories and somewhere in there you'll see either MSDOS Prompt or Command Prompt.
Alternatively, you can go to Start - Run and type 'command' into the box that appears.
Then follow the instructions given by Stephen above.
It's not a search: it shows the route data is taking between you and the remote site.
|
Thank you Wellie - will try.
Much appreciated!
HF
|
Not sure I've succeeded though. The 'run' method only yielded a few comments about milliseconds etc, didn't mean a lot to me. In Accessories I couldn't find a way.
Think I'd better be patient and wait till the site is back to normal again. But ta for your help ;)
|
The two methods I mentioned are just different ways of opening a DOS window on your PC. You won't get different results by using one or the other.
The MSDOS/Command prompt may be somewhere different on your menu: it's a shortcut which you can place anywhere you like. If it's not under accessories it might be near the end of your program menu.
I think the purpose of the traceroute is just to see where any bottlenecks occur. Each line is a computer that the data has to pass through between you and the web site. If the site is running OK, you won't learn anything useful by doing it.
|
Thanks again Wellie.
I am guessing therefore that, having tried what you suggested, there is not a problem with the data passing between me and the site.
Tis weird though that there seems to be a site problem, but this is not thrown up by your DOS method.
Sorry to sound completely ignoramus-like here (it's because I am) but I DO really and sincerely appreciate your help.
Thank you.
HF
|
I tried 3 tracerts since Stephen's request.
The first tracert results were all 110ms plus.
The second and third tests brought this message up "request timed out", repeated about 10 times.
I remember that the site has been getting slower and slower every day from 10 July onwards when Stephen put up the following message:
www.honestjohn.co.uk/forum/post/index.htm?f=2&t=14...5"... Having said all this, I did do a big data retrieval this morning of a database at around 11.30am - so that would have wallopped the server for a few minutes. Next time I will reprioritise the job so that it doesn't kill the server so much. However, that would have only been for a short while and you guys are reporting slowdowns right now. .... " <<
Is it remotely possible that the work Stephen did on 10 July has somehow inadvertently caused a fault somewhere that has caused the engine to seize up? (Did he remember to replace the oil-drain plug?"
|
|
I just did a couple more tests. This morning, my lowest result is 127ms, and the highest 672.
|
|
|
|
|
|
