Description of problem:
After an upgrade to F10, I have to bind the rfcomm devices manually by hand:
#: rfcomm bind rfcomm0
Version-Release number of selected component (if applicable):
Define rfcomm devices and restart bluetooth. Dev devices will not be created by default.
Steps to Reproduce:
1. Edit /etc/bluetooth/rfcomm.conf
2. Restart /etc/init.d/bluetooth
3. Check if /dev/rfcomm0 exists.
Devices are not created.
Devices should be created.
I also have got the same problem on my F10 machines.
Sample how to bind manually for all users suffering from this, might be possible to integrate this into the bluetooth init script.
#: /usr/bin/rfcomm bind rfcomm0
The problem ist, that there is no "rfcomm bind all" command in the init-script.
Fedora 9 had this command in its init-script.
Does anybody know why this i removed?
Created attachment 332632 [details]
Bastien, this mirrors what the script looked-like in F-9. Any reason not to apply it to F-10 and devel branches of the bluetooth package?
Split it off in a separate -compat sub-package. Could you please test this build, and let us know whether it works for you:
Thanks Bastian, tested your split packages and it worked. Did not see the new rfcomm init script right in the start, but then all did go fine.
From my side you can close this to be fixed in the next releases.
in Rawhide as of Oct 13th, 2009 there is again (or still) the "rfcomm bind all" missing. I have to manually do this to get the as "bind yes" configured bluetooth devices. (e.g. /dev/rfcomm0 only appears once I have entered "rfcomm bind all")
The startup script of bluetooth looks a lot differnt now, so the patch being part of the description above might not be relevant any more.
Michael, did you install the "bluez-compat" package? That looks to be the intended use now.
Thank you John!
Yes I did not install the "bluez-compat" package. This seems to fix the issue (after enabling rfcomm service).
Why is bluez-compat not installed by default? The template file /etc/bluetooth/rfcomm.conf does imply that "bind yes" would automatically bind at startup. So probably a comment mentioning that "bluez-compat" is needed for this would be helpful. Currently it is not following the "principle of least astonishment" ;-)
Anyways, thank you!
*** Bug 529027 has been marked as a duplicate of this bug. ***