(PDB-5672) Periodically analyze reports table #4004
Conversation
|
|
|
The test failure is a NPE in the new test. Perhaps the data it is looking for does not exist in PG 14? |
rbrw
force-pushed
the
pdb-5672-analyze-parent-partitions
branch
from
2311b05
to
daecbcf
Compare
rbrw
force-pushed
the
pdb-5672-analyze-parent-partitions
branch
from
daecbcf
to
9d7a7ae
Compare
rbrw
force-pushed
the
pdb-5672-analyze-parent-partitions
branch
from
9d7a7ae
to
d925940
Compare
| (defn analyze-partitioned-tables [db shutdown-for-ex] | ||
| ;; This assertion isn't a requirement, but it makes it obvious that | ||
| ;; the function is next.jdbc "safe", e.g. no nested tranactions. | ||
| (assert (not (sql/db-find-connection db))) |
There was a problem hiding this comment.
I don't think we should leave this assert in because it throws an Error (not an Exception) and this runs inside a ScheduledThreadPoolExecutor. If we fail to catch the Error ourselves, it will cause us to stop analyzing entirely.
There was a problem hiding this comment.
Right -- was planning to change that to "something else" before we finished, but may just remove it.
rbrw
force-pushed
the
pdb-5672-analyze-parent-partitions
branch
from
d925940
to
8d1f620
Compare
postgres (up to at least 16) never analyzes partitioned table parents Nearly fixed in 14, but removed before release: https://www.postgresql.org/about/news/postgresql-14-beta-1-released-2213/ https://www.postgresql.org/about/news/postgresql-14-rc-1-released-2309/ Assumes the analysis runs quickly enough to avoid needing any enforced serialization, and to avoid (with the current single threaded executor) ever delaying gc enough to matter.
rbrw
force-pushed
the
pdb-5672-analyze-parent-partitions
branch
from
8d1f620
to
af3348d
Compare
No description provided.