Read errors "No such file or directory 11054 invalid magic number 0 in tcache" and "Reader 11000 (4291737)

The Read process stopped due to error. The following messages are seen in the event log:

Error    2009-06-05 09:17:09.484000 2560 2636 Reader: No such file or directory 11054 - invalid magic number 0 in tcache  (from ORCL) [module ord]
Error    2009-06-05 09:17:09.484000 2560 2636 Reader: 11000 - sp_ordr failed: cant read tran cache  (from ORCL) [module ord]
Error    2009-06-05 09:17:09.500000 2560 2636 Reader stopped: Internal error encountered; cannot continue  (from ORCL)

The *tcache* files have been corrupted.

The *tcache* located in $SP_SYS_VARDIR/state directory have been corrupted. There are two files and they are also called transaction cache files. The workaround to have Read running again is:

a. Shutdown SharePlex
b. Rename the files tcache.a and tcache.b
c. Restart SharePlex. This should result in  new tcache.a and tcache.b files getting created.

As for the reasons the *tcache* and other cache files can get corrupted, there could be many reasons. The tcache files are related to transactions and are in the state of updation from time to time. If there is any glitch in Operating System or Shareplex, they can sometimes be corrupted, though this is rare. If they do get corrupted, then the normal running of Read is impacted. The workaround can cause loss of first open transaction in Capture queue. It may be worth to try and identiry the tables involved in transactions in Capture queue. At least they can be synchronized at a later time using means such as compare, etc.

If incurring out of sync on the tables is not an easy option, this can be brought to the notice of Support so that any other option can be sought, though it is usually very rare to get past such corruption without trying the workaround suggested above.