ADSM-L

Re: [ADSM-L] TSM Recovery log is pinning since upgrade to 5.5.5.0 code

2011-05-13 08:04:43
Subject: Re: [ADSM-L] TSM Recovery log is pinning since upgrade to 5.5.5.0 code
From: Richard Rhodes <rrhodes AT FIRSTENERGYCORP DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 13 May 2011 07:57:17 -0400
>
>
> Hi Richard!
> The log only contains all transactions (since the last database backup)
> when running in rollforward mode. When running in normal mode, only
> uncommitted transactions are written to the log. As soon as a
> transaction is completed, the transaction is committed to the database
> and thus removed from the log.

We're probably saying the same thing ;-) . . . here is how
I understand the log to work.

My understanding is that the log is a logical sequential-circular file
with a head  (where new trans are written) and tail(the oldest
uncommitted tran). As trans are created they are written
to the head (uncommitted).  As trans complete (commited),
commits are written to the log.   Thus tsm knows what to do after a
crash - needing everything from head to tail to put the db
back into a consistent state.  When replaying the log committed trans can
be ignored,
and uncommitted trans can be rolled back.
The space used by committed trans
between the head and tail is _not_ reusable.  Usable space
is what is in front of the head up to the tail. As the oldest
tran commits, the tail is pulled forward.  If any one  tran remains
uncommitted long enough for the head to come all the way around
and but into the tail . . . you filled the log.  In a sense,
committed trans are like expired files on tape that need
reclamation before the space is reusable.

I thought the difference going to roll-forward mode is that
the log tail is pinned at the last backup.  The tail is not
moved up until a backup has completed.   While is normal mode
the tail  moves forward as the oldest tran commits.

In a sense, the log is always pinned, but the pinning tran is always
changing as trans commit.  So I see logpin _problem_ as being caused by
two issues:
1)  Some long running (slow) tran keeping the tail pinned.
2)  The rate new trans are added and move the head ever forward toward the
tail.

That's my understanding how it works . . . which could be very wrong!

Rick




-----------------------------------------
The information contained in this message is intended only for the
personal and confidential use of the recipient(s) named above. If
the reader of this message is not the intended recipient or an
agent responsible for delivering it to the intended recipient, you
are hereby notified that you have received this document in error
and that any review, dissemination, distribution, or copying of
this message is strictly prohibited. If you have received this
communication in error, please notify us immediately, and delete
the original message.

<Prev in Thread] Current Thread [Next in Thread>