Database is not Starting after was added new BI Files

[rrojo7229]

Hi,

Please, could you give a tip?

Database is not starting UP...despite is not writing any ERROR message at database.lg.

The BI reached the limit last night and this morning the DB was shutdown and was added new BI Files.

But now when we try Startup the DB - keep hanging up in that normal Server message:

PROGRESS Version 9.1E02 as of Fri Jul 29 21:20:23 EDT 005
09:40:13 BROKER 0: Multi-user session begin. (333)
09:41:06 BROKER 0: Begin Physical Redo Phase at 256 . (5326)

That message still hanging UP since 1 hour ago...

What can I do?

Thanks in advance for any suggestion.

 

[Casper]

Wait, the database is in its redo phase. This can take quite some time to finish.

Is this large bi file growth expected behavior? How big was the bi file when the database crashed/was stopped?

Edit: any activity on the system?

what os?

Casper.

 

[richieasir]

redo phase means?

thanks in advance.

:jump:wishing u happy new year 2009:lol:

 

[rrojo7229]

Hi Casper,

Thank you for your quickly reply.

Actually, when the last BI variable got the Limit LINUX Size it was written these ERROR MESSAGES at log:

00:52:05 SYSTEM ERROR: Attempted to exceed maximum size on file /archive/bifiles/package.b3. (9452)
00:52:05 ** Insufficient disk space to extend the before-image file. (854)

Is there no way to try speed this process? Because since we tried the Multi-User session is taken 2 hours.

In tha last 5 minutes has this message, showing it had moved on.

Wed Dec 31 09:40:13 2008 09:40:13 BROKER 0: Multi-user session begin. (333) 09:41:06 BROKER 0: Begin Physical Redo Phase at 256 . (5326) 11:32:54 BROKER 0: Physical Redo Phase Completed at blk 454079 off 8137 upd 2067. (7161) 11:32:54 BROKER 0: Begin Physical Undo 2 transactions at block 454080 offset 211 (7163) 11:33:01 BROKER 0: Physical Undo Phase Completed at 452575 . (5331) 11:33:01 BROKER 0: Begin Logical Undo Phase, 1 incomplete transactions are being backed out. (7162)


I know, it depends of many things but How much time it will take from now on? Another 2 hours? or another half hour? It would be handfull if the process give an estimation for the time.


Progress 9.1E - Linux RedHat.

Regards,

 

[TomBascom]

There is no simple answer to "how long will it take".

Some bad thing happened which led to your bi file growing beyond the size limit. That caused the db to crash. Now the transaction which caused the bi growth is being undone. Undoing the TRX can take 2x or 3x the bi space that the original transaction was taking. It can also take 2x or 3x as long (or longer).

Unless you know what caused the bi growth in the first place and have a good idea how long it took to get to where it crashed you really have no good way to predict how long the recovery will take.

This is also one of those areas where a weak IO subsystem (like RAID5) really, really hurts.

 

[doom1701]

Did the DB come up?

When this happened to us a few months ago, on a fast system with *very* fast IO, it took almost an hour to roll back. A 2GB+ transaction is a hefty bit of data. Unfortunately, I never found out what caused it, but now I monitor bi size and I get an alert if the variable BI extent ever grows.