Darac Marjal <mailingl...@darac.org.uk> writes: > On Tue, Feb 24, 2015 at 11:23:08AM +0000, Sharon Kimble wrote: >> I have just bought a new computer,a Lenovo 050 desktop. Its working >> reasonably okay except for the occasional crashes. For instance, I've >> had it a fortnight now and its longest uptime has been just over 3 >> days, even though its on all the time! >> >> It has crashed when I've been using it, and its crashed overnight >> whilst I've been asleep! There does not seem to be any pattern to >> it, or any noticeable trigger. >> >> When it crashes, the screen freezes, and both the usb keyboard and >> mouse freeze, and can not be used. >> >> In its most recent crash, it shows this in logwatch the following day >> >> --8<---------------cut here---------------start------------->8--- >> --------------------- Kernel Begin ------------------------ >> >> >> WARNING: Segmentation Faults in these executables >> Watchdog : 2 Time(s) > > You appear to be running a watchdog process, which is crashing. A > hardware watchdog will, generally speaking, work as follows: at boot, it > will do nothing, but when an application writes a specific command to > it, it will start a timer (30 seconds, 60 seconds, etc. Often > configurable). When the timer expires, the watchdog will cause the > computer to reset. However, when the application writes the command to > the watchdog again, the timer is reset back to the start. So, as long as > the watchdog process is alive, it is able to keep resetting the clock > and the machine works fine. But if the processes stops (because the > machine has hung, or the system is too busy), then the timer will > expire, the machine will reboot and you can get back to work. > Or in my case just freeze everything!
> In your case, though, it looks like the watchdog process is dying > because of a Segmentation Fault (see the full logs or a core file for > full details). At that point, even though the rest of the system is > running fine, no-one is resetting the watchdog clock. > >> >> WARNING: Kernel Errors Present >> INFO: recovery required on readonly filesyste ...: 1 Time(s) >> mce: [Hardware Error]: Machine check ...: 4 Time(s) > > You are seeing Machine Check Exceptions. These may or may not be a > problem. This is usually the CPU saying "I ran into an error". This > could be the cause of your Segmentation Fault, for example - the > watchdog executes an invalid command, the processor complains and the > watchdog receives a SegFault as a result. I suspect the difference > between this and a SIGILL (Illegal Instruction) is that, while the > instruction is technically valid, the data or the state of the processor > or something like that is resulting in an error. It IS possible that the > code is perfectly correct, but the processor is calculating the wrong > answer (because of a hardware error). So can I get round it by stopping or disabling the watchdog? I don't know what its doing and haven't deliberately installed it. It just seems to have appeared from somewhere! > >> >> ---------------------- Kernel End ------------------------- >> >> --------------------- Connections (secure-log) Begin >> ------------------------ >> >> New Users: >> glances (123) >> >> New Groups: >> glances (134) >> >> >> Changed password expiry for users: >> glances : 1 Time(s) >> >> **Unmatched Entries** >> lightdm: pam_ck_connector(lightdm:session): nox11 mode, ignoring PAM_TTY >> :0: 1 Time(s) >> systemd-logind: New seat seat0.: 1 Time(s) >> systemd-logind: Watching system buttons on /dev/input/event3 (Power >> Button): 1 Time(s) >> systemd-logind: Watching system buttons on /dev/input/event4 (Sleep >> Button): 1 Time(s) >> systemd-logind: Watching system buttons on /dev/input/event5 (Power >> Button): 1 Time(s) >> systemd-logind: Watching system buttons on /dev/input/event7 (Video >> Bus): 1 Time(s) >> usermod: change user 'glances' password: 1 Time(s) > > You've included this information. Did you not create the user "glances" > yourself? > No, the user "glances" was auto-created when I installed glances yesterday. Sharon. >> >> ---------------------- Connections (secure-log) End >> ------------------------- >> --8<---------------cut here---------------end--------------->8--- >> >> What should I - >> a) - install, to get a better idea of why its crashing? >> b) - do to stop/get round the crashing please? >> >> Thanks >> Sharon. -- A taste of linux = http://www.sharons.org.uk my git repo = https://bitbucket.org/boudiccas/dots TGmeds = http://www.tgmeds.org.uk Debian testing, fluxbox 1.3.6, emacs 24.4.1.0
signature.asc
Description: PGP signature
