Omnia loses connection after approximately three days


I have Turris Omnia with mSATA drive running OpenVPN server. I am regularly accessing the data from outside via VPN service. I was in holidays in last two weeks and my Omnia always lost the connection after approximately three days. There was no issue/maintenance in the network (my provider is init7, in Basel, Switzerland)

During that time, there was no single client on WIFI, there was only VOIP phone connected via cable. After three days, It was not possible to access Omnia from outside using the VPN service. Also the VOIP phone showed lost connection. My neighbour made hard reset (simply pull out the power cable) to get it back online. The same happened after three days for the second time.

Did somebody observed similar behaviour? Is this a “power saving” feature or some bug?

Unfortunately, at that time, I was not copying the /var/log/messages to the disk so I do have logs only after reboot.

Any idea how to prevent it or document it in better way if that happens in future again?

I uploaded figure form the connection stats, which shows the connection losses.

I’ve seen a few users where /tmp/ gets filled in time, and that paralyzes much of its function. Example: Pakon.db filling up tmp but filling it may certainly happen in different ways, at least in theory.

I would rather expect issue with Pakon.db in case of heavy usage and not when the router is almost not used.

1 Like

The two times that this happened to me was an issue related to Suricata. I haven’t had an issue in the last five months, I believe the turris guys fixed the code a few versions ago. I should mention that I was working with tech support and they recommended that I reflash the router with the latest medkit and set the router back up manually which I did. The problem was definitely tmp filling up.

That is what worked for me, ymmv.