Rainer Weikusat rweiku...@mobileactivedefense.com writes:
[...]
It is possible to hit the 'fd limit'
bug (with a single client) by running squid with a tight file descriptor
limit (eg, 64) and trying hard enough. In order to make for easier
debugging, I changed the TcpAcceptor/ AcceptLimiter
On 07/16/2014 06:46 PM, Amos Jeffries wrote:
On 17/07/2014 11:10 a.m., Alex Rousskov wrote:
On 07/16/2014 02:38 PM, Rainer Weikusat wrote:
Alex Rousskov rouss...@measurement-factory.com writes:
On 07/16/2014 11:11 AM, Rainer Weikusat wrote:
This is broken because it re-enables readable
NB: This occurred in the real world on some 'customer machine' using
3.3.12. Since the code is unchanged in 3.HEAD, I assume it is affected,
too.
The method mentioned in the subject is
void
Comm::TcpAcceptor::doAccept(int fd, void *data)
{
try {
debugs(5, 2, HERE New connection on
On 07/16/2014 11:11 AM, Rainer Weikusat wrote:
void
Comm::TcpAcceptor::doAccept(int fd, void *data)
{
try {
debugs(5, 2, HERE New connection on FD fd);
Must(isOpen(fd));
TcpAcceptor *afd = static_castTcpAcceptor*(data);
if (!okToAccept()) {
Alex Rousskov rouss...@measurement-factory.com writes:
On 07/16/2014 11:11 AM, Rainer Weikusat wrote:
void
Comm::TcpAcceptor::doAccept(int fd, void *data)
{
try {
debugs(5, 2, HERE New connection on FD fd);
Must(isOpen(fd));
TcpAcceptor *afd =
On 07/16/2014 02:38 PM, Rainer Weikusat wrote:
Alex Rousskov rouss...@measurement-factory.com writes:
On 07/16/2014 11:11 AM, Rainer Weikusat wrote:
This is broken because it re-enables readable notifications even if no
connection was accepted.
In other words, it re-enables a check for new
On 17/07/2014 11:10 a.m., Alex Rousskov wrote:
On 07/16/2014 02:38 PM, Rainer Weikusat wrote:
Alex Rousskov rouss...@measurement-factory.com writes:
On 07/16/2014 11:11 AM, Rainer Weikusat wrote:
This is broken because it re-enables readable notifications even if no
connection was accepted.