Removed SQL 2012 AV group leaves phantom processes consuming large amounts of resource

Question

Hi,
I have been experiencing performance degradation on some production servers that previously were part of a SQL 2012 High availability Group.
I recently used a small window of opportunity to test the HADRON groups on some of my SQL Servers. Unusually because of the licencing and hardware high spec, identically setup servers required I had to take some of our servers out of the web production
pool during a quiet period and use these for testing.
When the testing concluded, I removed all the HA SQL groups disabled, then removed the windows cluster and removed the cluster Feature from both of the servers and rebooted, but I see the following phantom queries.
() select table_id, item_guid, oplsn_fseqno, oplsn_bOffset, oplsn_slotid
from [<dbname>].[sys].[filetable_updates_2105058535] with (readpast)
order by table_id
A reboot resolved the issue initially and although unhappy without knowing the root cause I decided not to spend any further time investigating as believed it was resolved. But I noticed a major hit on performance on those specific server again after 2-3
weeks and the queries have returned, if this happens during peak period I will be short of capacity and unable to reboot.
How can I completely remove trace of this as these are obviously protected system tables?.... there are still the odd days when I wished I was still allowed to meddle in those system tables :)

Answer 1

I'd guess that it is enabled in the process somewhere then, because we have never used the filetable feature, only AlwaysON.
The query detailed above runs for each user database - soaking up CPU i.e. at a fairly quite period this servers CPU is running at almost double another identical spec server in its pool on our website.
Looking the procedure cache, historically it if you combined the separate plans it spends
90% of CPU on these queries.