Open esiaero opened 1 month ago
Curious if this kind of behavior has been encountered before?
I haven't come across this behavior myself, but if that swap causes things to work better, it might be worth changing in our codebase.
I'll look into this a little bit today and update here.
Versions: PostgreSQL 16.4
Extensions are all the head of their respective repos except pg_partman which is missing a minor metadata file update. Exact commits: pg_partman - de93f4a pg_cron - f5d1111 columnar - 47354c2 timeseries - ba218b8
Manual installation of timeseries on Amazon Linux 2.0
The installation and extension creation goes fine. However, running
make installcheck
for the regression tests causes a segfault on extension creation.At a glance, I noticed
pg_partman
creation was going throughcolumnar
hooks. If I make it so cascade installspg_partman
beforecolumnar
The regression tests run fine (only difference being the order of cascaded dependent extension creation).
Curious if this kind of behavior has been encountered before?