Microsoft
Knowledge Base Article - 185577
XADM: Error 1120 and 5000 Occurs When
Starting Information Store
This article was previously published under Q185577
SYMPTOMS
The Information Store service in Microsoft Exchange Server
may not start
and the Server Specific Error 4294966746 occurs. Events
1120 and 5000 are
recorded in the Event Log with MSExchangeIS as the source.
CAUSE
This behavior can occur when the Information Store stops
abnormally and
the Eseutil.exe hard repair option (/P) is run against
the Priv.edb file to
return it to a state in which it can be started. In addition,
the previous
transaction log files were deleted, as recommended in
the Eseutil dialog box,
but hard repair was not run against the Pub.edb file.
At this point, the Priv.edb file is returned to a consistent
state through the
repair process, but the Pub.edb file is not. A consistent
database does not
require the presence of previous transaction log files
to start, but an
inconsistent one does.
WORKAROUND
Use either of the following solutions to resolve this
behavior.
Remove Pub.edb
Remove the Pub.edb file. This removes all public folder
data on the server.
IMPORTANT: If you choose to remove the
Pub.edb file, rather than repair it,
do not remove the Edb.log file that was just created during
the previous
startup attempt.
That failure marks the Priv.edb file as inconsistent and
Priv.edb requires the
presence of its log file so that it can start successfully.
If the Edb.log file is removed, the only way to return
the Priv.edb file to a
consistent state is to run the "eseutil /p"
command again.
Run Eseutil /p
Run the "eseutil /p" command (without the quotation
marks) against the
Pub.edb file.
MORE INFORMATION
The symptoms described in this article also occur if Pub.edb
is repaired, but
Priv.edb is not.
IMPORTANT: Note that using the Eseutil's
hard repair functionality is a last
resort option. Using that option may result in data loss
if there is no current
backup or if circular logging is enabled, thus preventing
restoration of
Exchange data to the time of failure. First try a soft
recovery by using the
eseutil /r command. If this does not
work, try the eseutil /p command for
a hard repair; this may be the best current alternative.
In the application event log, a series of events similar
to the following
accompany this behavior:
A Microsoft Exchange Server database is considered consistent
only after it
Event ID: 100
Source: ESE97
Type: Information
Category: General
Description: MSExchangeIS ((<Process ID>) ) The
database engine
<version> started.
Event ID: 108
Source: ESE97
Type: Information
Category: Logging/Recovery
Description: MSExchangeIS ((<Process ID>) ) The
database engine is
initiating recovery steps.
Event ID: 109
Source: ESE97
Type: Information
Category: Logging/Recovery
Description: MSExchangeIS ((<Process ID>) ) The
database engine is
replaying log file <exchsrvr>\MDBDATA\edb.log.
Event ID: 110
Source: ESE97
Type: Information
Category: Logging/Recovery
Description: MSExchangeIS ((<Process ID>) ) The
database engine has
successfully completed recovery steps.
Event ID: 1120
Source: MSExchangeIS
Type: Error
Category: General
Description: Error Database is in inconsistent state initializing
the
Microsoft Exchange Server Information Store database.
Event ID: 5000
Source: MSExchangeIS
Type: Error
Category: General
Description: Unable to initialize the Microsoft Exchange
Information
Store service. Error Database is in inconsistent state.
has been shut down normally. At all other times, including
during normal
operation, there is a flag in the database marking it
as inconsistent. Thus if
the database service is terminated abnormally, Exchange
Server knows on
the next startup that something went wrong in the previous
session.
Exchange then initiates "soft recovery" steps
to back out or commit
necessary transactions to the database and restore its
integrity.
You can check a database for consistency with the following
command:
eseutil /mh <database path and name> | more
The State line in the screen output from this command
may contain
Consistent or Inconsistent.
NOTE: Exchange 2000 Service Pack 2 and
later, do not report the database
state as "Consistent" or "Inconsistent"
but as "Clean
Shutdown" or "Dirty
Shutdown." The meaning of "Clean
Shutdown" is the same as "Consistent",
and the meaning of "Dirty Shutdown" is the same
as "Inconsistent".
The information in this article applies to:
Microsoft Exchange Server 5.5
Last Reviewed: 6/5/2003 (2.0)
Keywords: kbprb KB185577
© 2003 Microsoft Corporation. All
rights reserved. Terms of use Security & Privacy Accessibility