Conversation
|
@yeo should And can we delete the one that's not used? |
|
As of e906c55 the duplicate file is now removed. This was there because at once upon a time the sh.in file was generated from the .tmpl file, but that is no longer the case and having both there is turning out to be a bad idea :) |
As of eac55e0, when persist mode is active, `nbd-client` doesn't exit. By changing the systemd service to simple type and passing `nofork` to `nbd-client` , behavior is consistent between persist and non-persist, allowing the service to work as expected for both configurations.
|
So I'm not sure I 100% understand the proposed change. Can you explain how the proposed change will work in both persisting and not-persisting mode, and why I should merge it? :) |
Thinking about this more, I don't think this is the right solution to the problem I'm attempting to solve. The problem is that when using persist mode, the systemd service hangs when starting, stuck in the "activating" state until it times out: I think the root cause is the persist mode changes in 5b2d384. Perhaps the solution is to fork, even in persist mode. |
|
Also, |
Right, that's what I was afraid of, but I assumed I must have been wrong :) [...]
We could do that, but then we should probably have a "force fork" option, so that we are guaranteed to be forking.
I created #201 for that. |
As of eac55e0, when persist mode is active,
nbd-clientdoesn't exit.By changing the systemd service to simple type and passing
noforktonbd-client, behavior is consistent between persist and non-persist, allowing the service to work as expected for both configurations.