xpra icon
Bug tracker and wiki

Opened 8 years ago

Closed 8 years ago

#111 closed enhancement (fixed)

Keep ssh tunnel alive

Reported by: ahuillet Owned by: ahuillet
Priority: major Milestone: 0.3
Component: client Version: 0.1.0
Keywords: Cc:

Description (last modified by Antoine Martin)

SSH through a NAT will often stop working after some idle time. The "real" solution is to add ServerAliveInterval to ssh_config.

We should either pass this to the ssh tunnel command, or implement a heartbeat mechanism.

As it stands, my Xpra sessions die after about 2h of idle.

Change History (5)

comment:1 Changed 8 years ago by ahuillet

-o ServerAliveInterval=300 on ssh commandline solves the problem. I propose adding that by default.

Last edited 8 years ago by Antoine Martin (previous) (diff)

comment:2 Changed 8 years ago by Antoine Martin

Description: modified (diff)

My understanding of ssh is that specifying -o OPTION overrides whatever is in the user's config file, so we probably don't want to do that.
Some users may have a low alive-interval set, others not, it should really be up to them to set this value - and they can do that already, in a much more flexible way (per host, etc) via ~/.ssh/config

Maybe we should just enable pings in xpra with a reasonable delay (currently the pings are enabled only when displaying statistics and are sent every second)

Last edited 8 years ago by Antoine Martin (previous) (diff)

comment:3 Changed 8 years ago by Antoine Martin

Owner: changed from Antoine Martin to ahuillet
Status: newassigned

please close as 'wontfix' when the launcher is ready

comment:4 Changed 8 years ago by Antoine Martin

note: --enable-pings will send a ping every second, which will effectively keep the session alive, we may want to make the delay configurable and replace this with --ping-frequency=N

comment:5 Changed 8 years ago by Antoine Martin

Resolution: fixed
Status: assignedclosed

r1007 enables pings unconditionally with a delay of 20 seconds.
--enable-pings now only increases the frequency of pings to 1 per second.

This should detect dead connections.

Note: See TracTickets for help on using tickets.