Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 4 Feb 2008 20:49:03 +0300
From:      Lev Serebryakov <lev@FreeBSD.org>
To:        Ivan Voras <ivoras@freebsd.org>
Cc:        freebsd-stable@freebsd.org
Subject:   Re[2]: gjournal panic 7.0-RC1
Message-ID:  <1188949738.20080204204903@serebryakov.spb.ru>
In-Reply-To: <fo58j1$7hq$1@ger.gmane.org>
References:  <3aaaa3a0802030751w69ce59a9oeb869e3d87d92616@mail.gmail.com> <fo58j1$7hq$1@ger.gmane.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Hello, Ivan.
You wrote 3 =D1=84=D0=B5=D0=B2=D1=80=D0=B0=D0=BB=D1=8F 2008 =D0=B3., 23:35:=
44:

> If so, this is the same class of errors as ZFS (some would call it
> "tuning errors")
  Why this is ever possible on "stable" (I know, that "stable" doesn't
meand really "stable" these days, but at least it is not -CURRENT,
whcih can be experimental) release of server operating system?!
  I was courious (and upset) when read about ZFS panics, and now I'm
just courious and dissapointed.
  My humble opinion is, that release system (and I'm almost sure, that
this panic and ZFS panics will not be fixed whwn 7.0 becomes RELEASE
instead of RC1) CAN NOT PANIC on healthy hardware. AT ALL. NEVER. Ok,
every programmer (me too) makes mistakes. Software contains bugs.
Ok. But bugs are bugs. They MUST be fixed.
  And now we have special "tuning errors" which (correct me, if I
wrong, and I want to be wrong here!) will not be fixed. They will be
hidden under tuning guides and recomendations, and signs "Beware!
Dragons are here!" Hey, are FreeBSD experimental system or production
one?!
  Ok, ZFS (gjournal, whatever) need more resources (memory, journal
space, whatever) and system is mis-tuned and can not provide this
resource (in case of ZFS and 32bit systems it seems, that ADDRESS
SPACE is such resource). I understand, that provide GOOD service in such
conditions is not possible/valuable (especially in ZFS/32bit case).
But now system doesn't provide ANY service, it PANIC! Which can
disturb other services on this system, which can continue work even
without limited resource (for example, NAT can work even if HTTP proxy
is failed due to problems with gjournalled/ZFS cache).

 Ok, subsystem can not work efficiently. Degrade service. Swith to
"one 512 byte sector per second" speed mode. Disable all caches, doesn't
try to work with all features. Complain 10 times/second on console (and
dmesg buffer). Ok. But PANIC?! NO-NO-NO!!!

--=20
// Black Lion AKA Lev Serebryakov <lev@FreeBSD.org>




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1188949738.20080204204903>