Hi try and get the stack trace from the core ie mdb core.vold.24978 ::status $C $r
also run the same 3 mdb commands on the cpio core dump. also if you could extract some data from the truss log, ie a few hundred lines before the first SIGBUS Enda On 11/06/08 01:25, Krzys wrote: > THis is so bizare, I am unable to pass this problem. I though I had not > enough space on my hard drive (new one) so I replaced it with 72gb > drive, but still getting that bus error. Originally when I restarted my > server it did not want to boot, do I had to power it off and then back > on and it then booted up. But constantly I am getting this "Bus Error - > core dumped" > > anyway in my /var/crash I see hundreds of core.void files and 3 > core.cpio files. I would imagine core.cpio are the ones that are direct > result of what I am probably eperiencing. > > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24854 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24867 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24880 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24893 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24906 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24919 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24932 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24950 > -rw------- 1 root root 4126301 Nov 5 19:22 core.vold.24978 > drwxr-xr-x 3 root root 81408 Nov 5 20:06 . > -rw------- 1 root root 31351099 Nov 5 20:06 core.cpio.6208 > > > > On Wed, 5 Nov 2008, Enda O'Connor wrote: > >> Hi >> Looks ok, some mounts left over from pervious fail. >> In regards to swap and dump on zpool you can set them >> zfs set volsize=1G rootpool/dump >> zfs set volsize=1G rootpool/swap >> >> for instance, of course above are only an example of how to do it. >> or make the zvol doe rootpool/dump etc before lucreate, in which case >> it will take the swap and dump size you have preset. >> >> But I think we need to see the coredump/truss at this point to get an >> idea of where things went wrong. >> Enda >> >> On 11/05/08 15:38, Krzys wrote: >>> I did upgrade my U5 to U6 from DVD, went trough the upgrade process. >>> my file system is setup as follow: >>> [10:11:54] [EMAIL PROTECTED]: /root > df -h | egrep -v >>> "platform|sharefs|objfs|mnttab|proc|ctfs|devices|fd|nsr" >>> Filesystem size used avail capacity Mounted on >>> /dev/dsk/c1t0d0s0 16G 7.2G 8.4G 47% / >>> swap 8.3G 1.5M 8.3G 1% /etc/svc/volatile >>> /dev/dsk/c1t0d0s6 16G 8.7G 6.9G 56% /usr >>> /dev/dsk/c1t0d0s1 16G 2.5G 13G 17% /var >>> swap 8.5G 229M 8.3G 3% /tmp >>> swap 8.3G 40K 8.3G 1% /var/run >>> /dev/dsk/c1t0d0s7 78G 1.2G 76G 2% /export/home >>> rootpool 33G 19K 21G 1% /rootpool >>> rootpool/ROOT 33G 18K 21G 1% /rootpool/ROOT >>> rootpool/ROOT/zfsBE 33G 31M 21G 1% /.alt.tmp.b-UUb.mnt >>> /export/home 78G 1.2G 76G 2% >>> /.alt.tmp.b-UUb.mnt/export/home >>> /rootpool 21G 19K 21G 1% >>> /.alt.tmp.b-UUb.mnt/rootpool >>> /rootpool/ROOT 21G 18K 21G 1% >>> /.alt.tmp.b-UUb.mnt/rootpool/ROOT >>> swap 8.3G 0K 8.3G 0% >>> /.alt.tmp.b-UUb.mnt/var/run >>> swap 8.3G 0K 8.3G 0% >>> /.alt.tmp.b-UUb.mnt/tmp >>> [10:12:00] [EMAIL PROTECTED]: /root > >>> >>> >>> so I have /, /usr, /var and /export/home on that primary disk. >>> Original disk is 140gb, this new one is only 36gb, but disk >>> utilization on that primary disk is much less utilized so easily >>> should fit on it. >>> >>> / 7.2GB >>> /usr 8.7GB >>> /var 2.5GB >>> /export/home 1.2GB >>> total space 19.6GB >>> I did notice that lucreate did alocate 8GB to SWAP and 4GB to DUMP >>> total space needed 31.6GB >>> seems like total available disk space on my disk should be 33.92GB >>> so its quite close as both numbers do approach. So to make sure I >>> will change disk for 72gb and will try again. I do not beleive that I >>> need to match my main disk size as 146gb as I am not using that much >>> disk space on it. But let me try this and it might be why I am >>> getting this problem... >>> >>> >>> >>> On Wed, 5 Nov 2008, Enda O'Connor wrote: >>> >>>> Hi Krzys >>>> Also some info on the actual system >>>> ie what was it upgraded to u6 from and how. >>>> and an idea of how the filesystems are laid out, ie is usr seperate >>>> from / and so on ( maybe a df -k ). Don't appear to have any zones >>>> installed, just to confirm. >>>> Enda >>>> >>>> On 11/05/08 14:07, Enda O'Connor wrote: >>>>> Hi >>>>> did you get a core dump? >>>>> would be nice to see the core file to get an idea of what dumped core, >>>>> might configure coreadm if not already done >>>>> run coreadm first, if the output looks like >>>>> >>>>> # coreadm >>>>> global core file pattern: /var/crash/core.%f.%p >>>>> global core file content: default >>>>> init core file pattern: core >>>>> init core file content: default >>>>> global core dumps: enabled >>>>> per-process core dumps: enabled >>>>> global setid core dumps: enabled >>>>> per-process setid core dumps: disabled >>>>> global core dump logging: enabled >>>>> >>>>> then all should be good, and cores should appear in /var/crash >>>>> >>>>> otherwise the following should configure coreadm: >>>>> coreadm -g /var/crash/core.%f.%p >>>>> coreadm -G all >>>>> coreadm -e global >>>>> coreadm -e per-process >>>>> >>>>> >>>>> coreadm -u to load the new settings without rebooting. >>>>> >>>>> also might need to set the size of the core dump via >>>>> ulimit -c unlimited >>>>> check ulimit -a first. >>>>> >>>>> then rerun test and check /var/crash for core dump. >>>>> >>>>> If that fails a truss via say truss -fae -o /tmp/truss.out lucreate >>>>> -c ufsBE -n zfsBE -p rootpool >>>>> >>>>> might give an indication, look for SIGBUS in the truss log >>>>> >>>>> NOTE, that you might want to reset the coreadm and ulimit for >>>>> coredumps after this, in order to not risk filling the system with >>>>> coredumps in the case of some utility coredumping in a loop say. >>>>> >>>>> >>>>> Enda -- Enda O'Connor x19781 Software Product Engineering Patch System Test : Ireland : x19781/353-1-8199718 _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
