October 22, 201510 yr unRAID OS Version:6.1.3 Description:Kernel panic in interrupt related to apic timer? (see attached syslog and screenshot of the panic dump). How to reproduce:This is tricky, I haven't found anything to specifically trigger the panic other than leaving the server running and the panic seems to occur within about a day. At least once it happened within minutes after booting. And since no one else is reporting the issue it seems it must be hardware specific. However, I *can* eliminate the panic and the tell-tale time sync error in the syslog by adding "noapic" the kernel boot parameters or by downgrading to 6.1.2 6.0.1. Expected results:run new unraid versions out-of-the-box on the hardware I bought from limetech. Well, technically I've upgraded RAM and installed a slightly different CPU than what was delivered, so maybe this is a stretch Actual results:kernel either has a bug or has dropped support for my apic hardware, reducing uptime to about a day on average with constant unclean shutdowns requiring lengthy parity syncs Other information:Downgrade to 6.1.2 6.0.1 eliminated the crash, seems to be related to linux kernel version changes in 6.1.x This is the only line I see in the syslog that seems to hint at anything related to it: Oct 19 19:32:23 Tower ntpd[1408]: kernel reports TIME_ERROR: 0x41: Clock Unsynchronized and that line no longer appears in the log when using the "noapic" kernel option or when running 6.1.2. Kernel panic. I have a serial cable on order so hopefully I can capture more than just this last single screen, taken with a camera pointed at a monitor I temporarily hooked up to the server. syslog.txt
October 30, 201510 yr Author I'm now at 3 days of uptime after downgrading to 6.0.1. I did have a single bit of one memory address appear as bad on *only* the first pass of an 18 hour memtest, but I've reserved that region with a memmap= on the kernel command line. I'm pretty well convinced this kernel panic was a bug/hardware incompatibility introduced in one of the unraid versions after 6.0.1. Gonna give it a few more days of uptime, then increment up to 6.1.0. The linux kernel version was updated in 6.1.0, so maybe the crash will show up there.
November 2, 201510 yr Author 6 days of uptime on 6.0.1. Stepping up to 6.1.0 to see what happens. tower-diagnostics-20151102-1637.zip
November 5, 201510 yr Author 3 days of uptime on 6.1.0 with no incidents. Stepping up to 6.1.1. tower-diagnostics-20151105-1726.zip
November 13, 201510 yr Author 7 days of uptime on 6.1.1, moving to 6.1.2. If previous behavior is consistent, I'll see a kernel panic within 48 hours. I've got the serial connection together and working, so I expect to be able to capture the full panic log this time around. tower-diagnostics-20151112-1912.zip
Archived
This topic is now archived and is closed to further replies.