Massive writes reported just after a new start

[hawihoney]

I usually shutdown my machines every night and restart them when needed. Yesterday, for the very first time, the UI showed massive write counts approx. 20 minutes after the restart. The system was no longer responsible, shutdown didn't work. I had to use the power button.

 

After a new restart the machine started a parity check and works as expected until now.

 

Any idea? I will attach screenhot and syslog. I did cut the syslog (127MB for 3 hours machine activity). The rest shows repeated lines.

 

Parity had green ball, the rest was sleeping.

 

Thanks in advance.

syslog.zip

[RobJ]

That's quite an interesting screen pic!  Of course, it's wrong, the drives were non-responsive so no massive writes.

 

You've suffered a severe system failure.  I didn't try to trace all of the drives, a bit too confusing, with all of the disk controllers it says you have, but at least some of the failed drives were connected to onboard ports.  First, with so many drives failing almost simultaneously and being dropped, it's obviously not the drives, so you can be reassured your data is safe, and probably no corruption of any kind either.  But it does look like something on the motherboard failed.  The first thing I would check is the fan on the bridge chipset, on the motherboard, perhaps it has stopped and the chipset overheated.

 

By the way, it looks like you are still running in an IDE emulation mode for your main onboard SATA ports!  It would be a little safer and a little faster to change that in the BIOS, preferably to AHCI, but at least a native SATA mode.  I'm rather sure that board supports AHCI.

 

If it's not the fan, I don't know what you can do.  You can try a memtest, but I don't think it's a memory problem, but something related to the PCI bus or other controllers on the motherboard.

 

And a last point, now that you are on v6, there's a better option than including the syslog (thanks for that!).  Attach the diagnostics file (Tools -> Diagnostics), which has the syslog plus a lot of other info.  It might be interesting to see an lspci result.  However, in this case, the diagnostics zip file may have been too big too, as you had to remove so much of the repeated garbage in the syslog.

[RobJ]

If you have a backup of your flash drive from just before this crash, preserve it!  And make a current backup of the flash drive.  You may need it if you have to start over, or migrate to new hardware.

 

I was afraid to say it too soon, but perhaps I should warn you, this may be end of life for that motherboard.  It's had a long and good life though.  Hopefully it can be rescued, but you probably should still consider retiring it, even if it begins to work again.

[hawihoney]

Thanks for your answers. The failing one is an old MD1500 from LimeTech -2008 I think. My second server, an MD 1510/ll is only one year older. Time to say goodbye to some parts.

 

I would like to replace both motherboards, CPUs, RAM and SATA controller to new hardware. I would like to use the old cases.

 

My problem, none of my friends, collegues and neighbours has any clue about hardware.

 

Need to find one nearby Cologne/Germany...

 

[BoHiCa]

I recently experienced a similar problem with a machine acting like it was possessed by a demented circus monkey. The problem was bad capacitors on the motherboard.  One had even "erupted" like a volcano. You can often tell if a capacitor has gone rogue by simply looking at them. (Sometimes they smell too!) Capacitors look like little cans (sometimes bare aluminum in color, sometimes plastic wrapped various colors) but they almost always look like little cans soldered to the printed circuit board (pcb, mainboard). You can visually inspect them as they very often will show signs of bulging or leaking. The full-metal capacitors usually (modern ones anyway) have stress etching on the top to keep them from literally exploding and this will be bulged out on the tops of the cans.  There was a series of bad capacitors that appeared in a lot of low and middle-tier boards from the early 2000's, but I still see failed capacitors on boards from the mid 2000's also.  To reduce the risk somewhat, look for "Japanese" capacitors on the specs of any boards or components you buy.  There are capacitors on controller cards, video cards, and in power-supplies too, not just main boards.  Even solid state doesn't always last forever!