1.6 KiB
SSH Timeouts and connection time
Over time, there have always been a few complains about SSH connection timeout errors and slow SSH connection startup. These especially popped up in the latest release even though no obvious code was changed.
As it turns out, increasing the value for ConnectTimeout in SSH does not actually only change the timeout after which an error is thrown, it is also used by some servers as a guideline for their response time.
E.g. if you specify a 10s timeout, some servers will always take 10s to respond.
This is of course not mentioned in any of the spec but is more of an implementation choice.
In the latest release this caused more errors as the timeout was set higher. It should also have affected many SSH connections basically since the release of XPipe. I don't know how many people have been affected by this, it heavily depends on which ssh server and configuration your server runs. It for example happens on proxmox instances and my AWS EC2 instances.
This release should fix all of these issues simple by not specifying a connect timeout at all. Great work there.
I would like to exchange a few words with whoever thought: A newly connected SSH client specified a 10s connect timeout, that means we can sit around idle for 9 seconds. That is a great idea. Load balancing my ass, this also happens if the system is completely idle.
Changes
Fixes
- Fix actions that shut down XPipe, e.g. automatic updates and debug mode, not correctly executing if it exited too fast
- Fix error about nonexistent logs directory on first startup
- Fix possible NullPointers when checking whether current SSH session has died