Since the last roon core update the roon core process has been crashing and restarting itself for no apparent reason every few days (currently I see it happened on the 12th, 15th and last night). It’s not a problem except when I play white noise for my baby at night and since the core does not resume any playing streams it becomes a problem
Running on ubuntu 17.10 - linux core 4.13.0-38-lowlatency #43-Ubuntu SMP PREEMPT
Looking at the systemd log it seems that every time the core crashes it co-insides with an ntp time update:
There is noting in the server or RAAT server logs about this crash (I can send you the logs but there really is nothing besides the standard stuff in there), it just starts a new log and the old one still shows the playing stream. The message above about network configuration change is a false warning triggered by a router ipv6 broadcast, the network itself does not change nor does it disconnect.
The systemd-timesyncd daemon only operates when network connectivity is available and will run when the network link changes. The problem is more likely network related. Does this coincide with a DHCP lease renewal?
That particular message shows up every hour, from what I see online it’s timesyncd running on a schedule.
That machine is setup with a static lease in my router and it doesn’t change. The networkd log shows that the IP was obtained on April 17th and that’s it:
-- Logs begin at Tue 2018-04-03 10:25:12 CEST, end at Sat 2018-04-21 13:06:54 CEST. --
Apr 17 12:33:15 core systemd[1]: Starting Network Service...
Apr 17 12:33:15 core systemd-networkd[337]: Enumeration completed
Apr 17 12:33:15 core systemd[1]: Started Network Service.
Apr 17 12:33:15 core systemd-networkd[337]: eth0: Renamed to enp2s0
Apr 17 12:33:15 core systemd-networkd[337]: enp2s0: IPv6 successfully enabled
Apr 17 12:33:15 core systemd-networkd[337]: eth0: Renamed to enp0s31f6
Apr 17 12:33:19 core systemd-networkd[337]: enp2s0: Gained carrier
Apr 17 12:33:19 core systemd-networkd[337]: enp2s0: DHCPv4 address 10.11.12.145/24 via 10.11.12.1
Apr 17 12:33:19 core systemd-networkd[337]: Not connected to system bus, ignoring transient hostname.
Apr 17 12:33:21 core systemd-networkd[337]: enp2s0: Gained IPv6LL
Apr 17 12:33:33 core systemd-networkd[337]: enp2s0: Configured
Ok, so I found the systemd-networkd DHCP lease and it does seem to renew every hour at minute 32 so it does look like the crash coincides with the renewal. The question is why does it not happen every hour? I am playing white noise for at least 12 hours nonstop every night.
What’s the DHCP lease set to on your router? It looks like you’re dropping the connection and that’s why it’s renewing so frequently. Also, you have IPV6 enabled but it’s only set for link local.
Default lease time on the router is 7200 seconds. The lease on the linux box shows T1 at 3600, T2 at 6300 and LIFETIME at 7200 so it all seems in order (extending the lease at T1 from original DHCP server). I do not see anything in the logs to suggest that the link is dropped or changed, I’m using stock ubuntu with no custom network configuration whatsoever.
Also, what I have found online suggests that the “Network configuration changed, trying to establish connection.” message is misleading
Any reason for setting a 2 hour lease? What’s the output from sudo systemctl status systemd-timesyncd? The last time my box was updated was following a reboot earlier this month, i.e. a network link change. I’m running Ubuntu 16.04 LTS (Xenial) and continue to use /etc/network/interfaces not /etc/systemd/network. But that’s where I’d look first.
The default lease on my router is 7200 and I have not changed it. I changed it for that box to 24 hours now but that would just mask the problem, eventually it would happen again.
The timesyncd log shows time synchronization every hour along with that network changed message, it stopped after I changed the DHCP lease to 24 hours.
However, I don’t see why a DHCP extension should be crashing roon but only sometimes. When it doesn’t crash it music plays just fine - no interrupts, so I don’t think there is a link reset or change or I would have been hearing dropouts every hour.
At least you have a workaround; I’d push the lease out further since you’re using a reserved IP.
However, the facts as I see them are systemd-timesyncdonly displays the message when there is a network link change and this doesn’t happen every time the lease is renewed. I don’t think it’s DHCP but rather systemd-networkd or possibly systemd-resolved as this could be a DNS problem.
That would be my line of enquiry but I’m not expert in this area.
Just a small correction, in my case, the network changed message and time sync message show up together every time there was a DHCP renewal.
In any case, even if the network was changing and the connection was dropping, I don’t think the roon core should be crashing without even a stack trace.
I wonder if the ntp daemon is the culprit here. Maybe if the clock is moved back some logic in roon sees a negative elapsed time and crashes? That is the only thing I can think of so far why the crash is not on every DHCP renewal.
Thank you for contacting support and I do apologize about the delay in getting back to you here. I just want to touch base with you and ask if you are still having the difficulties playing music due to the DHCP renewal or if you have resolved the issue? Have you updated your Roon Core to version 1.5 yet and can confirm you are still experiencing the issue with the new Core Update? Please let me know when possible.
