Keeps restarting

[dcorsus]

Hi Dirk,

The UPNP logging was already set to "ERRORS ONLY". Did you want me to chang e it to something else?

M

nope, that is very interesting. I need to do some code reading. Did you already restart the PI or is still spitting out errors? If it is still in that state can you set the UPNP logging level to verbose for 2 minutes (error comes out each minute) and post that log. After that, restart the PI.

Dirk

[madas]

Sorry - I had already killed it and restarted. I'll monitor for the next time it happens.

[dcorsus]

Sorry - I had already killed it and restarted. I'll monitor for the next time it happens.

No problem, I'll try to do some codereading tonight

[dcorsus]

Sorry - I had already killed it and restarted. I'll monitor for the next time it happens.

Problem in the new code. Can you try the attached.
Dirk

[madas]

Running now - will report back

[dcorsus]

Running now - will report back

I looked at your log, there definitely is something going on at around 4am when all players disappear and reappear.

Would be interesting to observe this over longer periods so we can see if there is some correlation between time-of-day and issues.

It doesn't look like a major issue but one item has already attracted my attention. The Music DB is made at 2am and generates tons of traffic. In at least 2 of your logs, you lose a player about 30 min post that which is the time when the PI will timeout a player because it hasn't received an alive message. So either the players has dropped the alive message, the network has dropped it or your VM shared port has, or the PI
On top of that, one of your older traces had this bizarre issue where the socket interrupts the PI with x bytes to read and when read, gives an error. If this misread is for example the alive message, it would explain why suddenly a player drops and re-appears.

Dirk

[madas]

Hi Dirk,

There are some maintenance tasks that happen overnight so i'm not to concerned about the 2am to 6am period. Primarily i was bothered when it kept restarting itself every few hours throughout the day. Even if I restarted everything after the maintenance period is over it would always still restart.

Its been pretty stable the last few days - but just a few minutes ago it crashed completely. I've sent you the log.

M

[dcorsus]

Hi Dirk,

There are some maintenance tasks that happen overnight so i'm not to concerned about the 2am to 6am period. Primarily i was bothered when it kept restarting itself every few hours throughout the day. Even if I restarted everything after the maintenance period is over it would always still restart.

Its been pretty stable the last few days - but just a few minutes ago it crashed completely. I've sent you the log.

M

The log shows you lost a single player at 8:11am, I don't see any logs that it crashed completely. What exactly did you mean with crashed completely? The PI restarted? HS restarted? The VM crashed?

Looking at the logs, I see you lose a player now and then, you had 1 occurrence on 1/17 @ 5:57AM where the PI would have restarted itself in the past and it now recovered.

Care to comment on the maintenance tasks you run at night? Are some parts of your setup restarted/rebooted? Perhaps your router is set to restart itself on a set time? If you had a second PC (no VM) you could run it in parallel and see if the problem is in the network or with the VM based setup.

Dirk

[madas]

Hi Dirk,

I connected to the HS server and there was a message that HSPI_Sonos.exe had crashed on the screen. It looks like HS had already spawned a new process all by itself.

I run a daily VM snapshot of the machines overnight. That causes a few second blip (twice) during the maintenance period. None of my infrastructure (firewall, switch, etc) do any sort of reboot.

You are correct regarding the restarts. I think you have solved that problem. I was seeing multiple restarts a day (sometimes multiple an hour) and since the last version you gave me I've been able to go many days without a problem.

M

[dcorsus]

Hi Dirk,

I connected to the HS server and there was a message that HSPI_Sonos.exe had crashed on the screen. It looks like HS had already spawned a new process all by itself.

Did you by any chance catch the crash dump? I think they are stored somewhere in a windows directory, seems to recall something minidump but you may have to set it. You would be the first to actually have the PI crash so all of this is adding to the mystique of what might be going on. Do you restrict the VM to a maximum memory utilization? If so, how much?

Dirk

[madas]

Hi Dirk,

I didn't catch it - if I see it again i'll get more details.

I don't set any VM resource limits. The machine is massively overpowered for what it currently does.

All in all we are in much better shape now than I was a few weeks ago.

M