XFOR: Correlation Table Cleanup Fails; Ct2.dat & Ct2.idx Grow to Large Size (191773)

SYMPTOMS

The Linkage 3.2 Directory Exchange (LDE) (or Exchange 5.5 Connector for PROFS or SNADS) correlation table (CT) maintenance routine, CTCLEANUP, may fail to purge the CT files of expired entries. The \linkage\tables\Ct2.dat and Ct2.idx files may continue to grow to a large size. Additionally, the following may be logged in the Linkage Administrator Log browser:

   1998/07/22 16:53:22-      LME-PROFS-MEXDIA(006f) 3
    00504:LME-PROFS-MEXDIA started >> stdmain(878)
   1998/07/22 16:55:00-      LME-PROFS-MEXDIA(006f) 3 40740:Removing
    correlation table records older than 7 day(s)  >> ct2(1768)
   1998/07/22 16:55:00-      LME-PROFS-DIAMEX(01cf) 3 40740:Removing
    correlation table records older than 7 day(s)  >> ct2(1768)

   **** Next 2 entries significant ****
   1998/07/22 16:55:00-      LME-PROFS-DIAMEX(01cf) 1 40720:Error
    {CT:160()} finding correlation table record  >> ct2(1078)
   1998/07/22 16:55:00-      LME-PROFS-DIAMEX(01cf) 1 40750:Error {I/O
    error has occurred} scanning correlation table records  >> ct2(1874)

   1998/07/22 16:55:00-      LME-PROFS-DIAMEX(01cf) 3 40741:Correlation
    table cleanup is complete: {Successful}  >> ct2(1792)
   1998/07/22 16:55:00-      LME-PROFS(021d) 3 10100:Scheduled event
    CTCLEANUP is starting (running time: 0 minutes, restart delay: 0
    minutes) >> control(760)
   1998/07/22 16:55:00-      CONTROL-SERVICE(022b) 3 10100:Scheduled event
    CTCLEANUP is starting (running time: 0 minutes, restart delay: 0
    minutes) >> control(760)
   1998/07/22 16:55:00-      LME-PROFS-MEXDIA(006f) 1 40720:Error
    {CT:160()} finding correlation table record  >> ct2(1078)
   1998/07/22 16:55:00-      LME-PROFS-MEXDIA(006f) 1 40750:Error {I/O
    error has occurred} scanning correlation table records  >> ct2(1874)
   1998/07/22 16:55:00-      LME-PROFS-MEXDIA(006f) 3 40741:Correlation
    table cleanup is complete: {Successful}  >> ct2(1792)
   1998/07/22 16:55:00-      LDE(014a) 3 10100:Scheduled event CTCLEANUP
    is starting (running time: 0 minutes, restart delay: 0 minutes) >>
    control(760)

NOTE: This may also be an issue for the Linkage 3.2 Exchange SNADS Connector, because CTCLEANUP is a routine common to that connector.

CAUSE

Two circumstances result in this problem:

When the Ct2.dat file is large, a time-out on index node locking can result in index corruption (Ct2.idx) and return an error.
The CTCLEANUP routine aborts processing if any error is encountered during cleanup (above error), resulting in old entries remaining in the CT and file size growth during normal message processing.

These errors do not prevent new entries from being added to the tables; thus the CT2 files continue to grow.

WORKAROUND

Obtain the hotfix mentioned below. After obtaining the hotfix, perform the following steps:

While the Linkage services are stopped, delete the file:
\linkage\tables\Ct2.idx.
Replace the Ctreestd.dll and Lscms.dll components with the hotfix versions in the Exchsrvr\Bin directory. You may find some older versions in the Exchsrvr\Connect\Exchconn directory. Rename those older versions.
Restart the Linkage LDE services.
Reschedule the CTCLEANUP execution time, or wait for the next regularly scheduled time.

For information about rescheduling the CTCLEANUP routine, please review the online "Linkage Message Exchange Help". The Help file's index includes the topic "Change Schedule for Cleanup Process" under "Correlation Tables".

NOTE: The Ct2.idx file will be rebuilt at startup, but entries may not be purged until 24 or more hours have elapsed. See MORE INFORMATION below. Also, after the fix is applied, the CTCLEANUP routine will attempt to rebuild the index if an error is detected during runtime, thus handling the corruption condition in the future.

STATUS

Microsoft has confirmed this to be a problem in Linkage Message Exchange version 3.2.

A supported fix is now available, but has not been fully regression-tested and should be applied only to systems experiencing this specific problem. Unless you are severely impacted by this specific problem, Microsoft recommends that you wait for the next service pack that contains this fix. Contact Microsoft Product Support Services for more information.

Microsoft has confirmed this to be a problem in Microsoft Exchange Server version 5.5.

A supported fix is now available, but has not been fully regression-tested and should be applied only to systems experiencing this specific problem. Unless you are severely impacted by this specific problem, Microsoft recommends that you wait for the next service pack that contains this fix. Contact Microsoft Product Support Services for more information.

MORE INFORMATION

The correlation table is a mechanism used to track message acknowledgments from OV/VM and SNADS. By default, the CT holds entries for seven days. If an entry is older than seven days, CTCLEANUP purges the entry from the table. All data for the CT is stored in the Ct2.dat file, and its index is Ct2.idx.

Connector message flow stops while CTCLEANUP is running; thus the CTCLEANUP routine is only scheduled to run once a day by default. Even if CTCLEANUP's schedule is modified to run more than once a day, the purge routine within CTCLEANUP will not run if the timestamp of the zero byte file \table\CT2 (no extension) is the current day.