Historical bug tracker for the Pacman package manager.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
FS#45351 - [pacman] pacman-key autostarts gpg-agent (which prevents unmounting during installation of Arch)
Attached to Project:
Pacman
Opened by Sebastian Pipping (sping) - Tuesday, 16 June 2015, 18:42 GMT
Last edited by Doug Newgard (Scimmia) - Friday, 17 July 2015, 19:01 GMT
Opened by Sebastian Pipping (sping) - Tuesday, 16 June 2015, 18:42 GMT
Last edited by Doug Newgard (Scimmia) - Friday, 17 July 2015, 19:01 GMT
|
DetailsHi!
The guide on installing from another distribution at [1] proposes running # pacman-key --init # pacman-key --populate archlinux from inside the Arch chroot. Fine. Both of these call GPG binaries internally. Okay. At some point, GPG started autostarting an instance of gpg-agent unless it is invoked with --no-autostart, an option missing in --help output but documented in man page of GPG. The trouble with starting gpg-agent is that the agent process outlives the call to pacman-key and causes trouble when unmounting the chroot stack since gpg-agent is still using files from inside the chroot. Especially in a context where calls to pacman-key are automated, one would have to a) find that very gpg-agent process (the host system may have another) and terminate it or b) prefix the call to pacman-key by something like "unshare --pid --fork" to have put it into a dedicated process ID namespasce and have it killed when pacman-key terminates, automatically. I am wondering, if you would be willing to do something like that from inside pacman-key, i.e. be willing to a) pass --no-autostart to calls of GPG or b) prefix calls to GPG with "unshare --pid --fork" or so inside pacman-key. I'm curious for your reply. Best, Sebastian [1] https://wiki.archlinux.org/index.php/Install_from_existing_Linux |
This task depends upon
chroot ... pacman-key --init
chroot ... pacman-key --populate archlinux
chroot ... pacman -S ....
To my understanding, using arch-chroot would create and destroy the mount stack for each call again.
It would do a lot more than needed.
Personally, I believe this would better be addressed in pacman rather than arch-install-scripts.
- in pacman rather than [in] arch-install-scripts.
+ in pacman rather than by using arch-chroot.
I start to understand that the docs may need a different fix than what would ease (my) automation.
arch-chroot sh -c 'pacman-key --init && pacman-key --populate archlinux && pacman -S ...'
On the universally case: Please help me understand the plus side of having a GPG deamon running after pacman-key terminated if it wasn't running already. I don't see it myself, yet.