MQ Series not recycling after sticky bit change.

[Marco Santerre]

Hello all,

Well, I got a stumper here. Let me give you the bit of history first, and then, I'll explain.

On Monday, we tried implementing a sticky bit so that the world-writeable would not be able to delete files on our MQ file systems. Now, we know that that's a no-no.. and fixed it in the following 5 minutes. The queue manager took about 30 minutes to recover but did.

On Sunday, we recycled the queue manager (weekly process) but the queue manager never came up. No messages were recorded anywhere. Starting the queue didn't even give an error message, just the prompt back.

Now, we have Service Guard, so we switch the package over to the secondary node, and tried starting it and same problem. No message and just the prompt back even in manual mode. We rebooted our primary node and tried starting MQ on it again and it worked.

Now, what gives? I figured if it'd be something that is memory related it'd have disappeared when we switched over to the secondary node, but it didn't? If it is something that is more file system based, a reboot of the server shouldn't have cleared it?

We're pretty sure that the sticky bit change caused some corruption but the queue manager recovered.

So, my question is.. can it be something related to the system? if so, what should I be looking at? Any insights would be appreciated.

(I understand that this is not IBM but this is also being looked at by IBM, but I'm trying to cover the O/S part of things, which is HP-UX)

[Jannik]

It could be that you need the sticky bit for starting up the Application (MQ), but then it is started it can run without.
Try it at least!