The problem is distinguishing the difference between the keep alive traffic and a long transfer. Either Mina needs a more advanced idle measurement or SSHD would need its own idle manager.
On Fri, Oct 27, 2017 at 11:08 AM Goldstein Lyor (JIRA) <j...@apache.org> wrote: > Goldstein Lyor created SSHD-782: > ----------------------------------- > > Summary: Consider implementing a "reverse" keep-alive > mechanism in the server > Key: SSHD-782 > URL: https://issues.apache.org/jira/browse/SSHD-782 > Project: MINA SSHD > Issue Type: Improvement > Affects Versions: 1.7.0 > Reporter: Goldstein Lyor > Assignee: Goldstein Lyor > > > The current code contains only a keep-alive mechanism from clients to > servers. We should consider implementing a "reverse" keep-alive mechanism > from servers to clients using {{SSH_MSG_IGNORE}} messages. This mechanism > should be along the lines of the {{ClientAliveInterval}} and > {{ClientAliveCountMax}} parameters described in [sshd_config| > https://linux.die.net/man/5/sshd_config], The mechanism can be used to > avoid read timeout on the *client* side (e.g. SSHD-781) > > > > -- > This message was sent by Atlassian JIRA > (v6.4.14#64029) >