Describe the bug
If beegfs-mgmt --init is run without systemd running, it will fail with the following error:
[root@test-beegfs8-mgmt beegfs]# /opt/beegfs/sbin/beegfs-mgmtd --init
PANIC: panicked at mgmtd/src/main.rs:60:6:
Logger initialization failed: No such file or directory (os error 2)
In the above case, this is being run in a chroot to set up an image that will later be sent to the management node.
With strace, we can see the failure:
...
socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC, 0) = 3
connect(3, {sa_family=AF_UNIX, sun_path="/run/systemd/journal/socket"}, 30) = -1 ENOENT (No such file or directory)
close(3) = 0
write(2, "PANIC: panicked at mgmtd/src/mai"..., 111PANIC: panicked at mgmtd/src/main.rs:60:6:
Logger initialization failed: No such file or directory (os error 2)) = 111
write(2, "\n", 1
) = 1
sigaltstack({ss_sp=NULL, ss_flags=SS_DISABLE, ss_size=8192}, NULL) = 0
munmap(0x7f690d569000, 12288) = 0
exit_group(101) = ?
+++ exited with 101 +++
Describe the system
Test system is BeeGFS 8.3 on Rocky 9.6.
To Reproduce
Steps to reproduce the behavior:
- Create an OS image in a tree. Mount /proc, /sys, /dev, and chroot to that tree.
- Run beegfs-mgmt --init
Expected behavior
It would be beneficial to be able to handle bootstrapping functions without systemd actually being up and running. Our workflow is to install and handle most initial component setup in the chroot, enable (but of course not start) systemd services, then push image to node. Would it be possible to support this, or does this fundamentally go against the current design pattern of BeeGFS?
Describe the bug
If
beegfs-mgmt --initis run without systemd running, it will fail with the following error:In the above case, this is being run in a chroot to set up an image that will later be sent to the management node.
With strace, we can see the failure:
Describe the system
Test system is BeeGFS 8.3 on Rocky 9.6.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
It would be beneficial to be able to handle bootstrapping functions without systemd actually being up and running. Our workflow is to install and handle most initial component setup in the chroot, enable (but of course not start) systemd services, then push image to node. Would it be possible to support this, or does this fundamentally go against the current design pattern of BeeGFS?