Skip to content

Clk/srch fixes#280

Open
chriskirby1 wants to merge 5 commits intomainfrom
clk/srch-fixes
Open

Clk/srch fixes#280
chriskirby1 wants to merge 5 commits intomainfrom
clk/srch-fixes

Conversation

@chriskirby1
Copy link
Contributor

No description provided.

@chriskirby1
Copy link
Contributor Author

retest

@chriskirby1 chriskirby1 force-pushed the clk/srch-fixes branch 2 times, most recently from 2b85759 to 4a1cdbd Compare January 30, 2026 14:32
@chriskirby1
Copy link
Contributor Author

retest

1 similar comment
@chriskirby1
Copy link
Contributor Author

retest

@chriskirby1
Copy link
Contributor Author

retest

2 similar comments
@chriskirby1
Copy link
Contributor Author

retest

@chriskirby1
Copy link
Contributor Author

retest

@chriskirby1
Copy link
Contributor Author

retest

Because the srch triggers are inherently async to the test,
we can't be sure they won't fire prematurely just because a
compact worker started running at an inconvenient time.

Make the trigger arming silent to avoid spurious test failures.
Move the trigger arming closer to the point of interest to
increase the chances that we're actually testing what we want.

Signed-off-by: Chris Kirby <ckirby@versity.com>
Print the first and last entries, the entry_nr and entry_bytes.

Signed-off-by: Chris Kirby <ckirby@versity.com>
If a set of tests is provided, loop over the entire set the
requested number of times. Start and stop any requested tracing
across the set boundary.

Signed-off-by: Chris Kirby <ckirby@versity.com>
It's possible for a srch compaction to collapse down to nothing
if given evenly paired create/delete entries. In this case, we
were emitting an empty block. This could cause problems for
search_sorted_file(), which assumes that every block it sees
has a valid first and last entry.

Fix this by keeping a temp entry and only emitting it if it differs
from the next entry in the block. Be sure to flush out a straggling
temp entry if we have one when we're done with the last block of
the merge.

Signed-off-by: Chris Kirby <ckirby@versity.com>
The "server error emptying freed" error was causing a
fence-and-reclaim test failure. In this case, the error
was -ENOLINK, which we should ignore for messaging purposes.

Signed-off-by: Chris Kirby <ckirby@versity.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant