Microsoft Exchange Server database engine splits a column into several parts, when it finds it too large and thus, the Long-values (LVs) are created. These structures are stored in a different B-Tree (binary tree) of the associated table, each with a unique long-value ID. Each of these LVs has a specific header, called LVROOT, consisting of information, like reference count and length of LV. In some cases, this LVROOT can get overwritten (usually because of a defective B-split operation), which eventually results into corruption of LV. This can cause Information Store to stop, errors accessing LV and/or database corruption. To repair the database, you should restore from a suitable backup. If no valid backup is present, you need to apply an Exchange Repair solution.
To illustrate such problems, suppose you attempt to perform offline defragmentation of an Exchange database. But the operation fails with an error message as below:
Operation terminated with error -1526 (JET_errLVCorrupted, Corruption encountered in long-value tree) after seconds
Further, when you execute eseutil /g command to check the integrity of the database, it reports the database as corrupted.
The above error occurs due to corruption in LV tree. This can exist in conjunction with orphaned LVs.
You should first check for the backup. If present and valid, restore from it. In other case, you need to implement the given sequence of steps to solve this issue:
1. Run eseutil /g with /v and /x parameters for detailed output and examine it for the cause (error, orphaned or corrupted)
2. If the LVs are corrupted, you can run eseutil /p command. But make sure you backup the database first as this causes corrupted data to delete
3. If the LVs are corrupted and orphaned both, you should execute eseutil /d to