Carfield Yim <carfi...@gmail.com> wrote: > HI, I would like to ask some time I cannot get a core dump in a > solaris box, any reason causing this behaviour?
On May 14, 2:55 am, hume.spamfil...@bofh.ca wrote:
> Carfield Yim <carfi...@gmail.com> wrote: > > HI, I would like to ask some time I cannot get a core dump in a > > solaris box, any reason causing this behaviour?
Carfield Yim wrote: > On May 14, 6:03 am, "Richard B. Gilbert" <rgilber...@comcast.net> > wrote: >> Carfield Yim wrote: >>> HI, I would like to ask some time I cannot get a core dump in a >>> solaris box, any reason causing this behaviour? >> A crash caused by a hardware error could make it impossible to get a >> core dump.
> It actually just a memory issues
Memory is hardware! If the attempt to dump core triggers another memory error, the system may crash without writing the entire core dump.
Carfield Yim wrote: > HI, I would like to ask some time I cannot get a core dump in a > solaris box, any reason causing this behaviour?
Running setuid and/or setgid.
--
"I'm a doctor, not a mechanic." Dr Leonard McCoy <mc...@ncc1701.starfleet.fed> "I'm a mechanic, not a doctor." Volker Borchert <v_borch...@despammed.com>
Carfield Yim wrote: > On May 14, 10:20 am, nelson.bens...@gmail.com wrote: >> core file generation can be controlled with coreadm(1m) as well
> Great, here is the output, is that mean core is disabled?
> global core file pattern: /var/core/core.%f.%p.%n.%u.%g.%t > init core file pattern: /var/core/core.%f.%p.%n.%u.%g.%t > global core dumps: disabled > per-process core dumps: enabled > global setid core dumps: disabled > per-process setid core dumps: disabled > global core dump logging: enabled
The setting for "init core file pattern" is not the default. Supposing it hasn't been overridden on its way down the process tree, any process that is using credentials that don't allow it to write to /var/core will not be able to take a per-process core dump. To quote from the coreadm man page:
| Ordinary per-process core files are created in mode 600 | under the credentials of the process.
On May 14, 6:56 pm, "Richard B. Gilbert" <rgilber...@comcast.net> wrote:
> Carfield Yim wrote: > > On May 14, 6:03 am, "Richard B. Gilbert" <rgilber...@comcast.net> > > wrote: > >> Carfield Yim wrote: > >>> HI, I would like to ask some time I cannot get a core dump in a > >>> solaris box, any reason causing this behaviour? > >> A crash caused by a hardware error could make it impossible to get a > >> core dump.
> > It actually just a memory issues
> Memory is hardware! If the attempt to dump core triggers another memory > error, the system may crash without writing the entire core dump.
you should be able to use gcore to create a core file of a running process as well, but i'm getting the impression that you're trying to trace a fault of some kind (bounds error of seg fault maybe? yeah speculation).
if the process in question is running inside a zone, from what i understand it usually write the core file to the /var/core (or where ever configured with coreadm) location in the *global* zone. i couldn't see which version of solaris or whether or not zones are in use so that may not help at all.
i know that in various types of code you can disable core file creation by setting RLIMIT_CORE to zero in a call to setrlimit(2) or equiv func
On May 15, 6:18 am, nelson.bens...@gmail.com wrote:
> you should be able to use gcore to create a core file of a running > process as well, but i'm getting the impression that you're trying to > trace a fault of some kind (bounds error of seg fault maybe? yeah > speculation).
yeah, we are, look like we need to change coreadm setup
