I've running Mint 18.3 XFCE 64/Vista dual booting on a Fujitsu Li2727 laptop for about 7 weeks. Generally running really well, but twice in the last week I've had some problems while running Mint. First time none of the other partitions would mount, second time the Updater applet had a red cross/couldn't determine status. Both times I rebooted, and both times the system would not reboot. It would get to the login screen, but would then show error messages and stall at a black screen. From there I was able to run Fsck in repair mode on the Mint partition (there were error messages advising this) and this sorted it. I could still boot Vista OK.

If/when it happens again I'll make extensive notes so that I can describe the problem/fix properly, but meantime any suggestions would be appreciated.

My own guess is that it's an HDD problem, but SMART is reporting no bad sectors at this stage. I've attached a copy of the latest SMART report. Any suggestions from this would also be appreciated.

Attached Thumbnails

 

Disk controller, cable, memory issue or maybe cpu?

See if you can get the OEM hard drive diag and run it or use a generic test.

Thanks for that. I can rule out CPU hopefully, as I upgraded the CPU between 1st and 2nd failure. There's no cable on this laptop, and I'm waiting for a couple of sticks of "new" ram anyway, so if it's RAM that should sort it. If it's the controller it's new HDD time anyway I guess.

Meantime I'll see whether I can track down an OEM diag tool. Can't remember the make, but luckily it's only 2 screws to remove on this laptop.

It's a Seagate, so I downloaded Sea Tools for Windows and ran it from Vista. Drive failed the short disk self test - no info just FAIL and suggestion to run repair from Seatools for DOS. Then the laptop went completely dead not even power lights. Something very not happy. Failing HDD wouldn't do that would it?

Backing up just now.

"laptop went completely dead"

power issue??

It is possible the hard drive could do that but I don't think so. You might be able to remove hard drive and use a usb flash to run.

Power issue seems possible, but since it was plugged in and had a fairly good battery, that if it is power, it's likely to be scrap time. Booted straight back no problem and I've been able to backup OK, so that's one thing. I have a spare HDD with Tahrpup/Vista on it which I think I'll run for a few days. If the shutdown issue repeats, that would rule out HDD at least.

Not the HDD. Same issue with the spare Pup one. Decided to change back to the original CPU, but when I took the cover off I noticed that one ram stick was twisted and discoloured. Swapped out that one for a spare, and now back on Mint. Meantime I've bought a spare Laptop - same model - off Ebay for £30 just in case. At least that's one advantage of working with obsolete hardware.

No further problems since I changed the ram. That leaves me thinking that the original issue may have been related too. That just leaves me with the failed HDD test. I suppose running the Seagate long test from DOS as they suggest wouldn't hurt. Apparently that will try to fix problems as part of the test anyway. Anybody got experience of running that on EXT4 partitions?

Updating in case this is ever useful to anyone else. The "power" crashes returned after a day or two. After a lot of trial and error, it appears to be a slightly odd overheating problem - odd in that it would work fine and reasonably cool for days and then opening some random web page, libreoffice doc or launching a regularly used programme would quickly spike temps through the roof and it would die before I could do anything about it. Finally found some info from someone else who had suffered this with this laptop/cpu combo (Intel Merom T7500) on windows and had sorted his by undervolting. After research suggested that undervolting is not beginner friendly in Linux, I dropped max freq down from 2.2 to 1.6GHz and now everything seems to be sweet. Laptop still runs pretty well at the lower freq. I spent best part of a day trying to make the lower max freq (set with cpufreq-utils) survive a reboot - it apparently gets overwritten at boot time by the ondemand daemon - I found the indicator-cpufreq applet which lets me set it with 2 clicks after booting. Happy. My best guess at the source of the problem is that the laptop cooling system (which is clean as a whistle, has new Arctic Silver MX4 properly applied and a Gelid pad on the GPU) is incapable of cooling the higher spec cpu or possibly gpu I suppose, enough at full load, possibly due to thermal lag with the heatsink design (longish copper tube conducting heat off to the fan, but that the laptop rarely calls for sustained full load so this is an intermittent condition. I may at some stage pursue undervolting and/or work out how to change the cut in temp for the fan, but meantime I'm happy with it as is.

The original question was about disc corruption. I didn't end up running any more tests on the HDD, but instead transferred it to my "new" spare laptop. The problem reappeared once more on the new machine, but has stayed away for over 3 weeks of daily use for basic browsing by my wife. It will probably die at some point, but since it has no important data on it that's fine.