https://bugzilla.suse.com/show_bug.cgi?id=1183148 https://bugzilla.suse.com/show_bug.cgi?id=1183148#c18 --- Comment #18 from Michael Hamilton <michael@actrix.gen.nz> --- (In reply to Samuel Cabrero from comment #15)

I have reproduced the issue but using systemd-networkd instead of wicked and disabling systemd-networkd-wait-online.service.

In this situation the network-online.target is reached as soon as systemd-networkd.service starts, so nmb.service is started before getting a DHCP IPv4 address triggering the 5 seconds sleep.

But I can't reproduce it when using wicked. I have prepared a samba build with increased debug and having a shorter sleep interval. Could you please test the packages from this OBS repository and attach the system journal:

1. sudo zypper ar https://download.opensuse.org/repositories/home:/scabrero:/bsc1183148/ openSUSE_Tumbleweed/home:scabrero:bsc1183148.repo 2. sudo zypper --gpg-auto-import-keys ref 3. sudo zypper up --allow-vendor-change 4. reboot 5. sudo journalctl --boot -oshort-precise | tee journal.txt 6. sudo systemd-analyze plot > graph.svg 6. Attach journal.txt and graph.svg

Thanks.

Attached as requested. Reproducing the situation using wicked may depend on delays due to specific hardware/driver configurations (I'm running a renovated old system: AMD FX-6300, 16GB, SSD boot and home, ASUS M5A97 EVO with an onboard r8169 Ethernet controller). To reproduce something like my experience with wicked, it might be possible to insert sleep/usleep into network startup scripts to simulate delays due to hardware/divers. Out of curiosity I changed my boot sequence so that smb/nmb (and spamd) are boot disabled and added an after-boot timer-unit that starts them via a Wants clause. That reduces the boot to sddm prompt by 6 seconds. I described that here: https://forums.opensuse.org/showthread.php/551179-TW-boot-times-vs-15-2-15-3... I reverted to stock behaviour before running the test requested above. -- You are receiving this mail because: You are on the CC list for the bug.