Right,
it actually is not about a problem incurred, but about a schema change to be implemented, for stable queue (spool) syscdr tables control_send_stxn, trg_send_stxn and trg_receive_stxn, in order to enable one of the 14.10.xC6 ER performance and spool enhancement features by which hopefully most if not all of those nasty send queue problems should be put into everyone's past.
This particular change is about new in-row spooling which would place any transactions smaller than 26kB into an in-row ctbufdata_inline lvarchar field rather than the traditional ctbufdata blob field, thus avoiding a lot of the usual sbspace spool overhead, at least with typical OLTP workloads.
Per
migration guide, an instance should have it's queues empty and ER caught up to current log position, i.e. no more outstanding transactions to be replicated to other serers, before embarking on upgrade, under which circumstances nothing would be lost from a cleanstart anyway.
The cleanstart is being mandated, with any upgrade from pre-14.10.xC6 to xC6 or newer, as it implicitly would drop and re-create said tables - in the desired new schema.
So nothing to be scared about, and rather your entry point to an even better ER ;-)
HTH,
Andreas
------------------------------
Andreas Legner
------------------------------
Original Message:
Sent: Mon October 11, 2021 06:15 AM
From: Vicente Salvador Cubedo
Subject: CDR RQM Error: `ctbufdata_inline` column missing for on-disk queue tables
cdr cleanstart
That's the way. If you attended tonIIUG virtual conference there were very good presentations about changes in FC6 related to ER and how to upgrade.
Vicente
------------------------------
Vicente Salvador Cubedo
------------------------------