httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Yasser <yarafa...@gmail.com>
Subject Re: [users@httpd] DNS lookup failure for: hostname
Date Thu, 23 Apr 2015 23:34:00 GMT
I was digging through the proxy_util.c code and found the following:
if (err != APR_SUCCESS) {
        return ap_proxyerror(r, HTTP_BAD_GATEWAY,
                             apr_pstrcat(p, "DNS lookup failure for: ",
                                         conn->hostname, NULL));
    }

Variable err is value from the function call - apr_sockaddr_info_get found
in sockaddr.c which has a bunch of logic and if anything goes wrong there,
it could be misinterpreted as DNS lookup failure. The method
apr_sockaddr_info_get looks complex. I will have to dig deep into this.


Thanks,
Yasser

On Thu, Apr 23, 2015 at 6:29 PM, Yasser <yarafatin@gmail.com> wrote:

> Hi Yann,
>
> Yes, I tried that and didnt see any difference in the behaviour
>
>
> Thanks,
> Yasser
>
> On Thu, Apr 23, 2015 at 4:24 PM, Yann Ylavic <ylavic.dev@gmail.com> wrote:
>
>> Hi Yasser,
>>
>> Coudn't it be that your DNS server/network is not following the load?
>> Can you still reproduce the failures with jbosshost in your /etc/hosts?
>>
>> Regards,
>> Yann.
>>
>>
>> On Thu, Apr 23, 2015 at 8:20 PM, Yasser <yarafatin@gmail.com> wrote:
>> > The test ran for two hours on 4/21/2015 between 8pm and 10:50pm.
>> > I got the SAR report generated using ksar tool and didnt find anything
>> wrong
>> > with the resource usage. Memory was almost full but not swapping
>> either. A
>> > lot of TCP connections which is expected at this load. CPU and IO looks
>> > normal. Please find attached ksar report.
>> >
>> >
>> > Thanks,
>> > Yasser
>> >
>> > On Wed, Apr 22, 2015 at 9:48 PM, Robert Webb <rwebb@ropeguru.com>
>> wrote:
>> >>
>> >> That is exactly what I was thinking.
>> >>
>> >> But on to your real problem....
>> >>
>> >> What do the apache server's resources look like when this starts
>> >> happening? Things like number of connections, tcp port utilization,
>> cpu,
>> >> I/O, etc...
>> >>
>> >> Makes me wonder if there isn't something being taxed and causing the
>> >> server's DNS lookups to time out.
>> >>
>> >>
>> >>
>> >> On Wed, 22 Apr 2015 21:44:18 -0400
>> >>  Yasser <yarafatin@gmail.com> wrote:
>> >>>
>> >>> Yes.  The Apache servers may sound redundant with the current setup,
>> but
>> >>> in the future it will be used for other purposes like proxying to
>> >>> different
>> >>> virtual ips,  php  hosting etc.
>> >>> I cannot change the enterprise suite at this point in time.
>> >>> So are you running through the F5, load balancing to the two Apache
>> >>> proxies, then load balancing with the apache proxies to your 10 back
>> end
>> >>> JBOSS servers?
>> >>>
>> >>>
>> >>> On Wed, 22 Apr 2015 16:21:29 -0400
>> >>> Yasser <yarafatin@gmail.com> wrote:
>> >>>
>> >>>> I am in the midst of performance testing a web application and I
get
>> >>>> these
>> >>>> errors in Apache logs
>> >>>>
>> >>>> [Sat Apr 18 03:01:01 2015] [error] [client xx.xx.xxx.253] proxy:
DNS
>> >>>> lookup
>> >>>> failure for: jboss-host-name returned by /some/url, referer:
>> >>>> http://f5hostname/context/URI
>> >>>>
>> >>>> It occurs only at high load levels. I have seen posts where this
>> error
>> >>>> occurs due to incorrect proxy configuration, but my case doesn't
fall
>> >>>> under
>> >>>> that I guess. At the very end of the test I get another error -
"All
>> >>>> workers are in error state". I think the first DNS error causes
all
>> >>>> workers
>> >>>> to shutdown.
>> >>>>
>> >>>> Can someone point where I could start looking? Please let me know
if
>> you
>> >>>> need more details than whats mentioned below.
>> >>>>
>> >>>> *More details:*
>> >>>>
>> >>>> Apache version: 2.2
>> >>>>
>> >>>> mod_cluster: 1.2.4.Final
>> >>>>
>> >>>> I have an F5 in front of two Apache servers. There are ten JBoss
>> servers
>> >>>> in
>> >>>> the balancer.
>> >>>>
>> >>>> Apache configuration snippet: (I am not using proxypass directive,
>> just
>> >>>> modcluster)
>> >>>>
>> >>>> <IfModule mod_proxy.c>
>> >>>>    ProxyRequests Off
>> >>>>    ProxyVia On</IfModule>
>> >>>> <IfModule mpm_worker_module>
>> >>>>    ServerLimit, StartServers and worker related attributes configured
>> >>>> here</IfModule>
>> >>>> ProxyPreserveHost OnMemManagerFile /var/cache/httpd
>> >>>> <VirtualHost *:1280>
>> >>>>    ManagerBalancerName some2 (this is the same usedin jboss using
>> >>>> proxylist)
>> >>>>    ........
>> >>>>
>> >>>> Logs:
>> >>>>
>> >>>> [Tue Apr 21 22:06:25 2015] [debug] mod_proxy_http.c(1973): proxy:
>> HTTP:
>> >>>> serving URL http://jbosshost:8080/uri
>> >>>>
>> >>>> [Tue Apr 21 22:06:25 2015] [debug] proxy_util.c(2011): proxy: HTTP:
>> has
>> >>>> acquired connection for (jbosshost)
>> >>>>
>> >>>> [Tue Apr 21 22:06:25 2015] [debug] proxy_util.c(2067): proxy:
>> connecting
>> >>>> http://jbosshost:8080/uri
>> >>>>
>> >>>> [Tue Apr 21 22:06:25 2015] [error] [client 10.xxx.xxx.xxx] proxy:
DNS
>> >>>> lookup failure for: jbosshost returned by /URI
>> >>>>
>> >>>> [Tue Apr 21 22:06:25 2015] [debug] proxy_util.c(2029): proxy: HTTP:
>> has
>> >>>> released connection for (jbosshost)
>> >>>>
>> >>>> OS:
>> >>>>
>> >>>> RHEL 6.5
>> >>>>
>> >>
>> >>
>> >
>> >
>> >
>> > ---------------------------------------------------------------------
>> > To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
>> > For additional commands, e-mail: users-help@httpd.apache.org
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
>> For additional commands, e-mail: users-help@httpd.apache.org
>>
>>
>

Mime
View raw message