tag:blogger.com,1999:blog-3144569876865681673.post8689095505604195822..comments2023-06-22T03:38:41.686-07:00Comments on Mark's Project Blog: How to keep your gnome-terminal from hanging in SSH connectionsMark T. Tomczakhttp://www.blogger.com/profile/00635435261857863001noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-3144569876865681673.post-2536121755299656062021-07-07T09:25:20.397-07:002021-07-07T09:25:20.397-07:00While you're in there, if you work with SSH-ba...While you're in there, if you work with SSH-based git repository hosting services, you probably want to look into ControlMaster auto with a 10s or so timeout for connection reuse. This can massively increase the odds of success for bulk-updates when behind certain WISPs which rate-limit SYN, etc. Also, hard-disable agent forwarding and only enable where explicitly needed, and view those cases with prejudicial suspicion as to why things were designed to require it.Phil Phttps://www.blogger.com/profile/03639902544065628008noreply@blogger.comtag:blogger.com,1999:blog-3144569876865681673.post-28849106411164307802021-07-07T09:22:41.390-07:002021-07-07T09:22:41.390-07:00This smells like a case of an overly aggressive co...This smells like a case of an overly aggressive consumer-grade NAT device with a low timeout on TCP state, the sort of thing which works for HTTP but not most other protocols. You also sometimes see this with remote proxy gateways from certain cloud providers.<br /><br />TCP level keepalives can work, but having SSH do protocol-level keepalives can help. Meanwhile, not enabling this increases the chance that you can let your laptop sleep and then later open it up and the SSH connections could still work. A pleasant bonus for LAN hosts.Phil Phttps://www.blogger.com/profile/03639902544065628008noreply@blogger.com