| Subject | Re: [firebird-support] page 2226658, page type 5 lock denied (216) |
|---|---|
| Author | Ann Harrison |
| Post date | 2012-09-27T15:08:56Z |
On Thu, Sep 27, 2012 at 5:04 AM, Nick Upson <nu@...> wrote:
that was recognized and your process lost. The problem was entirely in
memory, so no, I wouldn't worry about datbase corruption. In theory,
Firebird should never have an internal deadlock. You've just seen a
difference between theory and practice. As processors get faster and loads
get heavier, more cases that were overlooked previously turn out to be
possible in fact. Without a repeatable test or a core dump, these problems
are very hard to diagnose. So if it happens once, chalk it up to bad luck.
If it happens repeatedly, be prepared to work with the core developers.
Good luck,
Ann
[Non-text portions of this message have been removed]
>Page type 5 is a data page. What you ran into was an internal deadlock
> (firebird 2.1.5, centos) I got this message in the firebird log and the
> client (running locally) got error code - 335544336
> and subsequent retry of the same operation on the same data had no
> problems.This system runs 24/7 so I don't have a
> simple way to run gfix, backups are taken ok and they restore.
>
> Should I be concerned about db corruption or anything else?
>
that was recognized and your process lost. The problem was entirely in
memory, so no, I wouldn't worry about datbase corruption. In theory,
Firebird should never have an internal deadlock. You've just seen a
difference between theory and practice. As processors get faster and loads
get heavier, more cases that were overlooked previously turn out to be
possible in fact. Without a repeatable test or a core dump, these problems
are very hard to diagnose. So if it happens once, chalk it up to bad luck.
If it happens repeatedly, be prepared to work with the core developers.
Good luck,
Ann
[Non-text portions of this message have been removed]