This error is as old as i know the peruser. Few years ago we debuged this. Put on debug mode, tcpdump. All we could see is that client just dropped the connection then socket was somehwhere in between multiplexer and processor. The apache request log showed that the request was served.
As far i as we could tell, then this request is aborted browser request. And if it weren't (network glitch) then browser should retry the request transparently to the user. I think that it was what we saw from the tcpdump also. But it was a long time ago, so I don't remember. Anyways the reason why this error is not silenced is that nobody actually hasn't completely confirmed that is't safe to ignore these, but also nobody hasn't confirmed that it's causing problems (because it's very hard to pinpoint this error, since it occures very very rarely ). Also this error *never* occured then running apache benchmark with various settings and multiple simultations client computers. Janno On 4.12.2009 12:21, Lazy wrote: > 2009/12/4 KoGt2S - Webmaster / XTRM Host<[email protected]>: > >> lazy, putting debug the apache log I got this ... >> >> >> >> [Thu Dec 03 20:54:30 2009] [info] Shared memory session cache initialised >> >> [Thu Dec 03 20:54:30 2009] [info] Init: Initializing (virtual) servers for >> SSL >> >> [Thu Dec 03 20:54:30 2009] [info] mod_ssl/2.2.14 compiled against Server: >> Apache/2.2.14, Library: OpenSSL/0.9.8e-fips-rhel5 >> >> [Thu Dec 03 20:54:30 2009] [debug] mod_security2.c(413): SecServerSignature: >> Changed server signature to "Apache/2.2.3 (Unix)". >> >> [Thu Dec 03 20:54:30 2009] [notice] Apache/2.2.3 (Unix) mod_perl/2.0.4 >> Perl/v5.8.8 configured -- resuming normal operations >> >> [Thu Dec 03 20:54:30 2009] [info] Server built: Nov 30 2009 22:30:15 >> >> [Thu Dec 03 20:54:30 2009] [debug] peruser.c(2562): AcceptMutex: sysvsem >> (default: sysvsem) >> >> [Thu Dec 03 20:54:39 2009] [info] (107)Transport endpoint is not connected: >> apr_socket_addr_get(APR_REMOTE) >> >> [Thu Dec 03 20:54:39 2009] [warn] (9)Bad file descriptor: >> apr_socket_opt_set: (TCP_NODELAY) >> >> [Thu Dec 03 20:54:39 2009] [info] (9)Bad file descriptor: >> apr_socket_addr_get(APR_REMOTE) >> >> >> >> if it helps you? >> > not, much you have to recompile with #define MPM_PERUSER_DEBUG uncomented > > but i think I know what it can be, if I'm right it's just client > closing connection so it's harmles > > more usefull whould be a tcpdump port 80 and apache error.log from > some time period when this warnings occur > > >> >> >> Thanks, >> >> >> >> KoGt2S >> >> _______________________________________________ >> Peruser mailing list >> [email protected] >> http://www.telana.com/mailman/listinfo/peruser >> >> >> > _______________________________________________ > Peruser mailing list > [email protected] > http://www.telana.com/mailman/listinfo/peruser > _______________________________________________ Peruser mailing list [email protected] http://www.telana.com/mailman/listinfo/peruser
