I'm trying to stress SQLonly. The reason I have zero confidence on the cluster H/W we are using.

create SQL job, select count (*) from sysprocesses where blocked 0, if not 0, send you an Email. This may help you.

I got correct answer from Microsoft. You can't add timestamp or any calculate column to the replicate database without re-initialize the snapshot.

You need consider where is your current SQL binaries. On the cluster, SQL binaries need on the shared cluster drive.

We had same problem, too. Are you on SQL 7 ? If do, you need go to SQL2K. Most of this problem disppear on SQL2K.

did you run dbcc to check any possible data corruption

I think the reason of the failure is because some transaction did not get replciated correctly. Did you try sp_repldone ? Is this on SQL 7 ?

I did use sp_repladdcolumn and get this error. The exact syntatx: sp_repladdcolumn @source_object = 'Table_A' , @column = 'TA_TS' , @typetext = 'timestamp', @force_invalidate_snapshot = 0, @force_reinit_subscription =...

My experience with HTT enabled sometime do more harm than help on a heavy load environment. You may need try to turn of HTT.