[users@httpd] intermittent problem - site unavailable

[Zoe Ballz <zo...@zoeballz.tv>]

Hi there

Has anyone ever heard of this problem before? It may or may not be an apache problem.

One of my servers has been running perfectly happily for about year. I have made no changes to httpd.conf except for adding and removing sites from the virtualhost section.

The server is win2000pro and apache 1.3.31 running as an application (not a service). It has 2 nics, 1 connected to adsl, 1 to a cable modem. each nic is configured so the gateway is the appropriate connection. Apache listens to both connections and until 4 days ago, ran fine. The server is an AMD 550mhz and only gets about 200 hits per hour.

On another machine I run a program called SiteMonak. This tests the server (by checking DNS - doing a ping - and making a request for a page) every 5 mins, and setting alarms ringing if its not available. The 1st test it does is to the site URL (the test machine uses the cable connection) , so the connection goes out the cable (via Nat router) - round the net - back in the adsl and (via another Nat router) off to the server. The 2nd test requests from the local IP and runs directly through my LAN.

The problem:

Every day alarms start ringing about 10 times a day when the external connection (1st test) appears to drop out for between 5-120mins (DNS and ping tests pass - http test fails). The 2nd test confirms the server is still operating. No other internet connections on either of my lines appear to be affected in any way. SMTP and FTP connections taking the same route as test 1 through the same routers (but routed to different internal IPs) remain unaffected.

So this is what is confusing me:

1. Test 2 working when test 1 fails >>>>>>>>>>>>>>> Implies an internet or LAN problem.

2. FTP connections can still be made ok >>>>>>>>>>>>>>> Implies Apache server problem.

3. Other people from other sites still being able to connect to the server with no interruption even when test 1 is failing >>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Implies server is OK.

4. The fact that when someone from a different ISP connects, test 1 suddenly springs back to life again (most the time) >>>>>>>>>>>>>>> implies Apache server problem or incoming router problem.

5. Some people on same ISP as the outgoing cable connection cannot connect when test 1 is failing >>>>>>>>>>>>>>>> Implies problem with the ISP.

There are no errors in the apache error logs. Only a lack of records in the access logs when test 1 is failing.

Can Apache accidentally ignore requests from particular IP? Can Apache accidentally ignore requests on a particular virtual host? Could there be a buffer or a table which is filling up, and being reset when someone else connects to a different Vhost?

Both my ISPs say they have no known issues (well they would wouldn't they).

Any help would be appreciated.

Zoe

http://zoeballz.net - Website Hosting for the Transgendered Community

Re: [users@httpd] intermittent problem - site unavailable

[Luis Moreira <lu...@esinnovation.pt>]

Zoe,

    Do you have static or dynamic DNS ? I suppose static, since you talk 
about a cable modem.
    Anyway, if you ping, then the machine itself is OK, so it's either 
Apache or the router that does not forward requests on port 80.
    If at the time you ping, the HTTP test fails remotely, try to 
"telnet 127.0.0.1 80" on the server.
    Providing Apache alllows for the internal IP to be called, if  it is 
up, you will get your homepage. If not, dig deeper into Apache logs.
    If this works, and you are still unable to access the site remotely, 
check the router(s).

Luis

Zoe Ballz wrote:

> Hi there
>  
> Has anyone ever heard of this problem before? It may or may not be an 
> apache problem.
>  
> One of my servers has been running perfectly happily for about year. I 
> have made no changes to httpd.conf except for adding and removing 
> sites from the virtualhost section.
>  
> The server is win2000pro and apache 1.3.31 running as an application 
> (not a service). It has 2 nics, 1 connected to adsl, 1 to a cable 
> modem. each nic is configured so the gateway is the appropriate 
> connection. Apache listens to both connections and until 4 days ago, 
> ran fine. The server is an AMD 550mhz and only gets about 200 hits per 
> hour.
>  
> On another machine I run a program called SiteMonak. This tests the 
> server (by checking DNS - doing a ping - and making a request for a 
> page) every 5 mins, and setting alarms ringing if its not available. 
> The 1st test it does is to the site URL (the test machine uses the 
> cable connection) , so the connection goes out the cable (via Nat 
> router) - round the net - back in the adsl and (via another Nat 
> router) off to the server. The 2nd test requests from the local IP and 
> runs directly through my LAN.
>  
> The problem:
>  
> Every day alarms start ringing about 10 times a day when the external 
> connection (1st test) appears to drop out for between 5-120mins (DNS 
> and ping tests pass - http test fails). The 2nd test confirms the 
> server is still operating. No other internet connections on either of 
> my lines appear to be affected in any way. SMTP and FTP connections 
> taking the same route as test 1 through the same routers (but routed 
> to different internal IPs) remain unaffected.
>  
> So this is what is confusing me:
>  
> 1. Test 2 working when test 1 fails >>>>>>>>>>>>>>> Implies an 
> internet or LAN problem.
>  
> 2. FTP connections can still be made ok >>>>>>>>>>>>>>> Implies Apache 
> server problem.
>  
> 3. Other people from other sites still being able to connect to the 
> server with no interruption even when test 1 is failing 
> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Implies server is OK.
>  
> 4. The fact that when someone from a different ISP connects, test 1 
> suddenly springs back to life again (most the time) >>>>>>>>>>>>>>> 
> implies Apache server problem or incoming router problem.
>  
> 5. Some people on same ISP as the outgoing cable connection cannot 
> connect when test 1 is failing >>>>>>>>>>>>>>>> Implies problem with 
> the ISP.
>  
> There are no errors in the apache error logs. Only a lack of records 
> in the access logs when test 1 is failing.
>  
> Can Apache accidentally ignore requests from particular IP? Can Apache 
> accidentally ignore requests on a particular virtual host? Could there 
> be a buffer or a table which is filling up, and being reset when 
> someone else connects to a different Vhost?
>  
> Both my ISPs say they have no known issues (well they would wouldn't 
> they).
>  
> Any help would be appreciated.
>  
> Zoe
>  
> http://zoeballz.net - Website Hosting for the Transgendered Community