[systemd-devel] issue of failing to start Network Name Resolution
Hello, I am setting up the wifi for a customized Linux image. It uses systemd to manage the network. Right now, I am able to config the ip address and ping 8.8.8.8, but get "bad address" when ping google.ca. It is because the systemd-resolved service failed to start. systemctl status shows below info: Active: failed (Result: exit-code) Process: 254 ExecStart=/lib/systemd/systemd-resolved (code=exited, status=226/NAMESPACE). Main PID: 255 (code=exited, status=226/NAMESPACE) journalctl shows below info: systemd-resolved.service: Failed to set up mount namespacing: /run/systemd/unit-root/dev: No such file or directory systemd-resolved.service: Failed at step NAMESPACE spawning /lib/systemd/systemd-resolved: No such file or directory I guess the issue is that the files required for systemd-resolved to execute successfully are not there. But by googling I couldn't find how to set up these two files and how systemd-resolved sets up the name server. Has tried to configure the name server in /etc/systemd/resolved.conf, /etc/systemd/resolved.conf.d/dns.conf and /run/systemd/resolve/resolv.conf. Or add DNS=xxx in /etc/systemd/network/20-wireless-dynamic.network. None of them work. I am wondering if something is missing there or if any debugging methods can be used. Thanks! Regards, Crane
[systemd-devel] journalctl: switches & more with less
Would it be possible to use switches with the basic KEY=VALUE searches. For instance, journalctl -g "state changed" _SYSTEMD_UNIT=ModemManager.service + UNIT=ModemManager.service _PID=1 + OBJECT_SYSTEMD_UNIT=ModemManager.service _UID=0 + COREDUMP_UNIT=ModemManager.service removing the -g "state changed" shows when MM is crasing, but adding it filters all of that out. or for instance journalctl -b -1 -u ModemManager + COREDUMP_UNIT=ModemManager.service This specific example seems like it should not be necessary if I read the manual. But something about `+ COREDUMP_UNIT=name.service _UID=0 MESSAGE_ID=fc2e22bc6ee647b6b90729ab34a250b1` does not show coredumps. The MESSAGE_ID looks extraneous. At any rate, it would be nice to be able to add to --unit switches. It would also be nice to be able to do something like journalctl -u ModemManager -k -b -1 I realize that I can do journalctl _SYSTEMD_UNIT=ModemManager.service + UNIT=ModemManager.service _PID=1 + OBJECT_SYSTEMD_UNIT=ModemManager.service _UID=0 + COREDUMP_UNIT=ModemManager.service + _TRANSPORT=kernel -b -1 But that is kind of long winded for something that feels like it should be intuitively easy. Version at hand is systemd 252 (252.12-1~deb12u1) +PAM +AUDIT +SELINUX +APPARMOR +IMA +SMACK +SECCOMP +GCRYPT -GNUTLS +OPENSSL +ACL +BLKID +CURL +ELFUTILS +FIDO2 +IDN2 -IDN +IPTC +KMOD +LIBCRYPTSETUP +LIBFDISK +PCRE2 -PWQUALITY +P11KIT +QRENCODE +TPM2 +BZIP2 +LZ4 +XZ +ZLIB +ZSTD -BPF_FRAMEWORK -XKBCOMMON +UTMP +SYSVINIT default-hierarchy=unified
[systemd-devel] enabling cgroups v2
*ISSUE 1:* In ubuntu20, centos8 and almalinux8: *root@host:~# grep cgroup /proc/filesystemsnodev cgroupnodev cgroup2* In centos7: *root@centos:~# grep cgroup /proc/filesystemsnodev cgroup* In my centos 7, command '*grep cgroup /proc/filesystems'* does not output "nodev cgroup2". Does this mean cgroups v2 is not supported in it and cannot be enabled in it? *ISSUE 2:* In my ubuntu20, centos8 and alma8 server, I enable cgroups v2 by: 1. adding "systemd.unified_cgroup_hierarchy=1" to GRUB_CMDLINE_LINUX in /etc/default/grub, 2. executing the command "grub2-mkconfig" or "update-grub" as root and then 3. reboot the server. Without reboot, cgroups v2 does not come into effect when i check using the below command: *root@host:~# stat -fc %T /sys/fs/cgroup/tmpfs* Is there a way this can be done without a reboot? I am looking for a way to enable cgroups v2 which should n't need a reboot.