Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

clarify TODO for whitebox disable_wal=1 in db_crashtest.py #11665

Closed
wants to merge 2 commits into from
Closed
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 14 additions & 4 deletions tools/db_crashtest.py
Original file line number Diff line number Diff line change
Expand Up @@ -318,10 +318,20 @@ def is_direct_io_supported(dbname):
}

whitebox_default_params = {
# TODO: enable this once we figure out how to adjust kill odds for WAL-
# disabled runs, and either (1) separate full `db_stress` runs out of
# whitebox crash or (2) support verification at end of `db_stress` runs
# that ran with WAL disabled.
# TODO: enable this at random once we figure out two things. First, we need
# to ensure the kill odds in WAL-disabled runs result in regular crashing
# before the fifteen minute timeout. When WAL is disabled there are very few
# calls to write functions since writes to SST files are buffered and other
# writes (e.g., MANIFEST) are infrequent. Crashing in reasonable time might
# currently assume killpoints in write functions are reached frequently.
#
# Second, we need to make sure disabling WAL actually makes a difference for
# `db_stress` runs where it is disabled. In particular, the full `db_stress`
# runs in the second half of the whitebox crash test need to run with
# `-reopen > 0` in order for `-wal_disabled` to make any difference.
# However, `-wal_disabled=1` is currently unsupported with any reopening.
# This issue could be resolved by adding the reopening support, or by moving
# the full `db_stress` runs out of the whitebox crash test.
"disable_wal": 0,
"duration": 10000,
"log2_keys_per_lock": 10,
Expand Down