Re: Puzzling System Action

OK, a bit more details...

1) With RAID you have protection against loss of one drive per parity set.
If more than one drive fails, or controller card goes down, then you are
out.
Obviously, mirroring provides higher level of protection - unit, controller
or bus level, depending on setup. But it comes at a greater cost - all
hardware should be duplicated.
So RAID and mirroring are two alternative protection strategies - you
choose what better suits your environment.
(RAID and mirroring can be mixed - you can have RAID-protected and mirrored
units in the same ASP. This creates greater granularity for choosing best
protection.)

4 and 5) I believe, entry sequence number in  journal is upto 2 billion+
(4-byte signed integer).
For system-managed journals system is supposed to do whatever maintenance
is needed automatically. I am not sure whether it will be reset or will
just wrap - I think it will be reset by system.
Occasionally, QDBJRN needs manual attention - I remember a system with
QDNJRNxxx journal receiver hitting 2GB limit. System freezed up and would
not work until CHGJRN was done manually. But it was clearly not designed to
be so and should not happen.

6) SMAPP is System-Managed Access Path Protection - system takes care of
access path which were not explicitly journalled.
You control SMAPP from EDTRCYAP display. This feature is described in
Backup and recovery Guide.

Hope this helps...



Best regards
    Alexey Pytel



Dave Mahadevan <mahadevan@fuse.net> on 04/27/99 11:34:12 AM

Please respond to MIDRANGE-L@midrange.com

To:   midrangel <MIDRANGE-L@midrange.com>
cc:    (bcc: Alexei Pytel/Rochester/IBM)
Subject:  Re: Puzzling System Action





Alexey:

> 1) When you start mirroring on ASP, which contains RAIDed drives, system
> will consider RAIDed drives already proctected and will leave them alone.
> Only non-RAID protected drives will be paired.
> In other words, if you have ASP containing only RAID-protected drives,
> starting mirroring for them will not change anything.
> If you really want mirrored protection, stop RAID-protection for drives
in
> ASP, and then start mirroring.
>
Ok I understand.  Stupid question:  How am I protected when the
subsystem goes down?  I  switched off one of the subsystems.   Jobs
accessing libraries in the RAID freezes and the system reports a missing
disk with A6xx0244.  Interesting behaviour.   I guess the protection is
only on the unit level within the subsystem.  Pardon my ignorance, why
cant RAID systems be mirrored?

> 4 and 5) QDBJRN (in QRECOVERY) is an internal system journal, which
> protects system dictionary files (QADBxxx). System takes care of this
> journal automatically, so normally user does not need to do anything
about
> it. If you want to reset sequence numbers, try to do it in a restricted
> state.
>
Is resetting really necessary?  How high can the sequence numbers go?

> 6) When SMAPP is active (EDTRCYAP), system may write internal - invisible
-
> entries to journals. These entries record changes to not-journaled access
> paths.These entries cannot be accessed by user, but they do consume
> sequence numbers.
>
>
You lost me here.   Maybe a little more explanation or a manual
reference?  Thanks.

> Best regards
>     Alexey Pytel
>

--
Thank You.

Regards

Dave Mahadevan.. mailto:mahadevan@fuse.net


+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to
MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator:
david@midrange.com
+---



+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---