Dennis Clarke wrote:
Ben, I'm reading this and calling up Cory Omand also who is the resident
Apache2 guru here at Blastwave. We know that the worker model has been
tweaked here to cooincide with work from the CoolStack. I am sure that
Stefan Teleman will also have some insights. Regardless of what Apache
may be doing it does not explain the userland process "wedgeing" the
whole zone and then the Global zone.
Let me look at this closely and .. I want to reproduce this here.
I have snv_59 here on Sparc. Will that suffice ?
I also have an Opteron machine that needs to be brought up to snv_59.
Let me know what I can do to help to duplicate the issue and then
perhaps track down the root cause.
The issues I'm having is on snv_43 on Opteron. As of yet I have no
reason to believe that this is solved on snv_56 (which I'm migrating all
our systems to) and while there is a sendfilev fix in snv_57 I'm not
certain that applies here.
Unfortunately this issue is not reproducible. So little information is
available when it happens that I can't isolate commonality by which to
formulate a method to reproduce. Frankly, if I could reproduce it I
could fix it... and thats the kicker.
The big pictures issue for me is still the basic: why can a process end
up in this state? and why can't I do something about it other than reboot?
Thanks to the mdb tips offered in this thread I'll hopefully have some
better information next time I encounter the problem.
If you guys come up with anything I'll be really interested. I
appreciate you having a look Dennis. ;)
benr.
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
