Title: Re: [Hardhats-members] Re: A troublesome and concerning error.....

Ah, but locks can be maintained at the user process level and still be immune to the lock holder being killed with a -9!  A GT.M process awaiting.a lock expects to be woken up when the lock is available, but will wake itself up every so often if it isn't woken up.  That way, even if the process holding the lock is terminated without a chance to clean up, the lock is recovered.

-- Bhaskar

------Original Message------
From: Greg Woodhouse
To: hardhats-members@lists.sourceforge.net
ReplyTo: hardhats-members@lists.sourceforge.net
Sent: Nov 3, 2005 12:13 PM
Subject: Re: [Hardhats-members] Re: A troublesome and concerning error.....


On Nov 3, 2005, at 4:01 AM, Bhaskar, KS wrote:

> Again, this seems consistent with the semantics of the JOB 
> command.  I don't see where killing a process with SIGKILL (9) 
> drops you back to the direct mode.

Neither do I (directly). But it still raises the question of whether 
an application can be written in such a way that an unrecoverable 
error could drop a user into programmer mode.
>
> To me, what is interesting is that killing the process owning the 
> lock doesn't allow the process waiting for the lock to eventually 
> get the lock, and you have to hit ^C.  In GT.M, the process waiting 
> for the lock eventually gets the lock, as in the example below.
>

Yeah...I noticed that. I suspect the issue is one of how locks are 
maintained. If it is done at the kernel level, then (e.g., lockf) the 
locks should be freed. But if they are maintained in application 
code, then there could be a problem.

===
Gregory Woodhouse
[EMAIL PROTECTED]

"Einstein was a giant. He had his head in the clouds and his feet on 
the ground."
--Richard P. Feynman




-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for free - -and be entered to win a 42" plasma tv or your very own
Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
Hardhats-members mailing list
Hardhats-members@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/hardhats-members


--------------------------
Sent from my BlackBerry Wireless Handheld

Reply via email to