[Gluster-users] Handling dropped packets from client

[Douglas Tan]

Hi,

My deployment environment has a firewall that removes idle connections and drops packets from these connections. How should I handle this situation for GlusterFS clients (native and NFS)? The docs mention the following options:

http://download.gluster.com/pub/gluster/glusterfs/3.2/Documentation/AG/html/chap-Administration_Guide-Managing_Volumes.html#sect-Administration_Guide-Managing_Volumes-Tuning

* network.frame-timeout
* network.ping-timeout

http://gluster.org/community/documentation//index.php/Translators_options#protocol.2Fclient

* transport-timeout
* ping-timeout

I'm trying to obtain the idle connection period that is configured for the firewall from the client that actually manages it but I'm unsure of which option should be the ones that I need to tune. TCP keep alive can be configured on the server so that will be something else that I can modify, if required.