It did.
Seems like it's not absolutely random, 'cause the next crash's possibility is higher after a crash.
I started to fix every bug listed in /var/log/messages. Maybe this crash will go away when I'm finished. Maybe not
Next suspicious element:
Mar 11 08:01:42 geburah NetworkManager: <info> Activation (eth1) Stage 5 of 5 (IP Configure Commit) scheduled...
Mar 11 08:01:42 geburah NetworkManager: <info> Activation (eth1) Stage 4 of 5 (IP4 Configure Get) complete.
Mar 11 08:01:42 geburah NetworkManager: <info> Activation (eth1) Stage 5 of 5 (IP Configure Commit) started...
Mar 11 08:01:42 geburah avahi-daemon[1329]: Joining mDNS multicast group on interface eth1.IPv4 with address 192.168.2.103.
Mar 11 08:01:42 geburah avahi-daemon[1329]: New relevant interface eth1.IPv4 for mDNS.
Mar 11 08:01:42 geburah avahi-daemon[1329]: Registering new address record for 192.168.2.103 on eth1.IPv4.
Mar 11 08:01:43 geburah NetworkManager: <info> (eth1): device state change: 7 -> 8 (reason 0)
Mar 11 08:01:43 geburah NetworkManager: <info> Policy set 'Auto default' (eth1) as default for routing and DNS.
Mar 11 08:01:43 geburah NetworkManager: <info> Activation (eth1) successful, device activated.
Mar 11 08:01:43 geburah NetworkManager: <info> Activation (eth1) Stage 5 of 5 (IP Configure Commit) complete.
This appears before every crash, but sometimes goes through without any problem.
Mostly it's just before "kernel: ...."-like lines (booting).
Possible origin of error:
NetworkManager: supplicant_interface_acquire: assertion `mgr_state == NM_SUPPLICANT_MANAGER_STATE_IDLE' failed
I don't really know what this does but the error could be some statuschange in network.
Any suggestions?