Hi,

My C application is linking to libmemcached. While testing my code for the
network cable pullout case, I see that all of the processes using
libmemcached set/get calls are stuck for quite a while (upto 15 minutes!).
The stack trace indicates that all of them are stuck inside the close() call
on the socket fd. I presume the set/get call gets a fatal error and hence
decides to close the socket fd before coming out with an error.

In general, how do people deal with such network cable down cases? Are there
some settings within libmemcached which can help to come out earlier in such
failure cases?

Regards,
Nikhils

Reply via email to