[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: trunk proxy_http failures
Thanks, Yann, this helped me pin the problem down further:
- with disablereuse=on everything works fine
- with ttl=1 the problem is still there
and then:
- with mpm_worker, the problem also disappears (no disable/ttl needed)
These tests were running since the dawn of h2 time and are still running in 2.4.x. Since the problem also goes away on worker, this looks like a new problem with mpm_event and connection closing (keepalive)?
Anything I can do to give you good data? Some preferred log settings?
-Stefan
> Am 19.09.2018 um 16:32 schrieb Yann Ylavic <ylavic.dev@xxxxxxxxx>:
>
> On Wed, Sep 19, 2018 at 4:10 PM Stefan Eissing
> <stefan.eissing@xxxxxxxxxxxxx> wrote:
>>
>> <Proxy "balancer://https-local">
>> BalancerMember "https://127.0.0.1:SUBST_PORT_HTTPS_SUBST" hcmethod=GET hcuri=/
>
> Does disablereuse=on help here?
>
>> </Proxy>
>> ...
>> ProxyPass "/proxy" "balancer://http-local"
>> ProxyPassReverse "/proxy" "balancer://http-local"
>>
>> Typical log:
> ...
>> [Wed Sep 19 14:02:58.038790 2018] [proxy:debug] [pid 35806:tid 123145377832960] proxy_util.c(2630): [client 127.0.0.1:64231] AH00947: connected /files/data-1m to 127.0.0.1:12346
>> [Wed Sep 19 14:03:03.041506 2018] [proxy_http:error] [pid 35806:tid 123145377832960] (70014)End of file found: [client 127.0.0.1:64231] AH01102: error reading status line from remote server 127.0.0.1:12346
>
> I suspect mod_proxy is reusing a connection already close by 127.0.0.1:12346.
> If you know the KeepAliveTimeout configured there, setting a ttl= to a
> lower value on the BalancerMember could be less radical than
> disablereuse...
>
> HTH,
> Yann.