| > PCM is still not running at this customer site. We seem to be getting
> closer!
> More questions:
> (1) Can a Hughes Data Systems (HDS) X Windows terminal be used?
To my knowledge we have never tested the any of the X-windows
interfaces or action routines with this kind of terminal. If it's
a reasonably good x-server implementation then it should work fine. You
really want to have a 3button mouse though.
> (2) How are reboots of the managed machines done from PCM?
Typically using the connect interface.
> Can they be done remotely?
Thats one of the goals of the product. You can do anything from PCM
that you could if the console was a VTxxx video-terminal.
> Can the entire VAXcluster be rebooted by a single command?
You would have to use the DIALOG interface to accomplish a cluster
reboot. There is no "automagic" command to reboot an entire cluster.
> (3) Again, the customer is concerned about reliability of the
> system. What happens if the system that the PCM is on becomes
> unavailable (they don't like a single point of failure)?
I would inform them that as long as there is only one console there is
*ALWAYS* a single point of failure.
> What happens if one managed node connection breaks?
> Do you keep the old consoles available for a backup?
You can always keep a VTxxx on a roll-around cart to accomodate this
situation.
> (4) The PCM User's Guide describes to do with VAX 8500/8700/8800
> and 8830/8840. No mention of 8820 -- which is what they
> have. Does the discussion for the 8830/8840 apply?
What kind of 8820? If it's an 8820N then it is the same thing as an
8800. If it's an 8820P then the 8830/8840 discussion applies. Look
at the console and it's a Pro-console the it's an 8800.
Regards,
Dan
|
| >Hi,
>Customer is still asking this question: if the computer the
>PCM is running on hangs what happens to the VAXcluster?
>Occassionally we have found that if a console hangs there a potential
>that the computer it is connected to may crash (due to a problem of the
>OPCOM messages overflowing the buffer). Now the customer is concerned
>he will have to worry about all the nodes in his VAXcluster having
>this problem at the same time. Can someone comment on his concern?
It's not that big of a concern. Over the years OPCOM has had
many internal changes to prevent this from happening. There is no
longer "a buffer" that can be overflowed. Messages are kept in a
linked list. It would take a long time for OPCOM to exhaust non-page
dynamic storage pool on most systems. Surely if you had a node with
a small memory config that disallows having a decently size non-paged
storage pool it would happen sooner. Finally, it has been such a long
time that I have ever had this occur ( many years) that I don't
consider it an issue. To put it another way, a paper jam on a local
hardcopy console is much more likely to happen.
Regards,
Dan
Thanks.
|