Error -1018 (JET_errReadVerifyFailure) in Exchange Server 4.0, in Exchange Server 5.0, and in Exchange Server 5.5 (151789)

CAUSE

When you perform a transaction with the Jet database, the information store or the directory store writes the transaction to a transaction log file (Edb*.log in Mdbdata or Dsadata). The transaction is then committed to the Jet database. During this process, the Jet engine calculates the page's checksum value to be written, records it in the page header, and then requests that the file system writes the 4-KB page of data to the database on disk. In brief, the file system takes this call and uses Windows NT system services to forward this request to the corresponding hardware device driver to actually perform the write. The hardware device driver returns this information to the file system, which then returns it to the Jet engine. If the call is successful, Jet continues.

Faulty hardware or hardware device drivers may return SUCCESS to calls made to them before they actually perform the physical operation. When the actual physical operation takes place, however, an error occurs and the data is not written successfully as expected.

In certain database operations such as but not limited to Online Backup, the backup routine makes a call to the operating system to read a 4-KB page of data from the database on disk and write it to tape. Before committing the data returned from the operating system call to the tape, the online backup process compares the checksum value in the page header (recorded when this page was written to disk) to that being returned from the READ call. If the checksum values do not match, the JET database engine detects this and returns -1018 (JET_errReadVerifyFailure).

RESOLUTION

To resolve the problem, use one of the following methods.

Restore from a Backup

When you encounter the error, the recommended and most reliable way to recover the database is to restore from the last known good backup.

NOTES:

Even after you restore from a known good backup, however, the -1018 errors may appear again unless the root causes of the physical data write problems are resolved.

The Windows NT system log should be reviewed for errors reported by the hard disks and hard disk controllers as well at the NT Fault Tolerant disk subsystem (FTDISK). In particular, hardware-related errors such as SCSI time-out errors, memory errors, or drive corruption errors may appear. These errors should be dealt with and resolved before the database is restored to this server.

Even if no hardware-related errors appear in the Windows NT system log, a hardware problem may exist on your computer.

Microsoft recommendeds that you contact your hardware vendor for further updates regarding your hardware.

Use Edbutil.exe or Eseutil.exe

An alternate way to repair the bad page of data is to use Edbutil.exe for Exchange Server 4.0 and 5.0, or Eseutil.exe for Exchange Server 5.5.

IMPORTANT: Make a full offline file backup of the database and log files before running this process. This includes file-level backups of the Exchsrvr\Dsadata and Exchsrvr\Mdbdata folders. This ensures that there is a way to go back if Edbutil/eseutil fails.

Edbutil.exe (and Eseutil.exe) is a low-level Jet database utility that defragments and checks the consistency of the database. There is also a repair option (Edbutil /d /r <dbname>, Eseutil /p <dbname>) that attempts to repair bad pages and remove them if the repair is unsuccessful. If a page is removed using the repair option, data loss will occur. A page in a database can contain messages, folders, tables, master indexes, and so forth. There is no way to tell what Edbutil or Eseutil has removed because Jet sees it as raw data. After repair has finished on the Jet database, run an offline defragmentation (Edbutil /D <dbname> or Eseutil /p <dbname>) to repair the space trees. Immediately after defragmentation, run Isinteg.exe in fix mode on the same database that Edbutil or Eseutil was run on. Isinteg cleans up any inconsistencies between the pointers from the information store to the actual data in the Jet database. If Isinteg is not run immediately following repair, corruption may occur in user mailboxes. Users will have messages that cannot be opened, incorrect read/unread counts, and so forth. After running Edbutil or Eseutil, make a FULL backup of the database that was repaired.

Refer to the Microsoft Exchange Server Administrator's Guide for more details on the use of Edbutil.exe and Isinteg.exe.

Troubleshoot the Problem

If you are unable to resolve the problem by restoring from a backup or using Edbutil.exe or Eseutil.exe, use the following steps to troubleshoot the problem. After performing each step, check to see if the problem is resolved.

Restart the computer. If you do not want to restart the computer at this time, stop and restart the information store and directory services instead. To do so:
1. In Control Panel, double-click Services.
2. Click Microsoft Exchange Directory, and then click Stop. If a dialog box appears stating that additional services will be stopped, note the services that will be stopped, and then click OK. The information store service should be included in the list of services that will be stopped.
3. Click Microsoft Exchange Directory, and then click Start.
4. Start any additional services that were stopped in step b, including the information store service.
Use the Chkdsk.exe utility to verify the integrity of your hard drives.
If the Exchsrvr folder is located on a compressed NTFS file system drive, uncompress the drive.
If write-back caching is enabled on a SCSI hard disk controller installed in your computer, verify that the cache is not volatile. Non-volatile caches are protected by a battery, Flash RAM, or some other method.
Disable any programs that keep files open for an extended period of time, such as anti-virus programs, file and disk management utilities, or backup programs. If disabling a program prevents the problem from occurring, verify that the program is completely compatible with Exchange Server before you enable it again.
Run Performance Optimizer (Perfwiz.exe) on the server.
Verify that the virtual memory settings in Windows NT Server are correct. To view the virtual memory settings, in Control Panel, double-click System, click the Performance tab, and then click Change.
If you are running any hard disk utilities obtained from your original equipment manufacturer (OEM), verify that you are running the most recent version of the programs.
Use the Performance Monitor counters associated with the following objects to detect potential problems:
- Cache
- Database
- LogicalDisk
- Memory
- MSExchangeIS
- Paging File
- PhysicalDisk
- Process
- Server Work Queues
- Server
- System
- Thread