You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is not a bug, it's designed by GPDB, there are 128 ao segfiles at most while 0 segfile is used for utility mode.
limit 128 files because TupleID for ao table is composed by [segfileno, rowno, flags] and the size is same with a TupleID for heap tables. It's a tradeoff of concurrent write process and rowno for ao tables.
But for now, the above problem occurs when the concurrency reaches 32. @avamingli
shmiwy
changed the title
[Bug] Insert into a ao relation with 127 concurrent transactions failled
[Bug] Insert into a ao relation with 32 concurrent transactions failled
Sep 21, 2023
bug for now, the above problem occurs when the concurrency reaches 33. @avamingli
I guess the reason is GUC gp_appendonly_insert_files is changed to 4 by default, it's designed for parallel. Also a tradeoff between concurrent processes and segfiles for parallel.
Show that GUC and set it to 1, have a check.
Cloudberry Database version
No response
What happened
setup
For each concurrency, just use the following query.
The system will report an error
What you think should happen instead
The system must not report errors
How to reproduce
rerun the query with 32 concurrent transactions
Operating System
centos7
Anything else
No response
Are you willing to submit PR?
Code of Conduct
The text was updated successfully, but these errors were encountered: