Update feature/perf again #6173
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Test that the event is correctly executed. Signed-off-by: Frediano Ziglio <[email protected]>
Signed-off-by: Frediano Ziglio <[email protected]>
Do not use ">" or other operators to compare Mtime.t, the value is intended to be unsigned and should be compared with Int64.unsigned_compare as Mtime functions do. Signed-off-by: Frediano Ziglio <[email protected]>
Signed-off-by: Frediano Ziglio <[email protected]>
The parameter was false only to support an internal usage, external users should always alert the thread loop. Signed-off-by: Frediano Ziglio <[email protected]>
- Do not wait huge amount of time if the queue is empty but use Delay.wait if possible; - Fix delete of periodic events. In case the event is processed it's removed from the queue. Previously remove_from_queue was not able to mark this event as removed; - Do not race between checking the first event and removing it. These 2 actions were done in 2 separate critical section, now they are done in a single one. Signed-off-by: Frediano Ziglio <[email protected]>
Potentially a periodic event can be cancelled while this is processed. Simulate this behavior removing the event inside the handler. This was fixed by previous commit. Signed-off-by: Frediano Ziglio <[email protected]>
Previously if the queue was empty and the loop thread was active the scheduler took quite some time to pick up the new event. Check that this is done in a timely fashion to avoid regressions in code. Signed-off-by: Frediano Ziglio <[email protected]>
Similar to test_empty test however the state of the loop should be different. Signed-off-by: Frediano Ziglio <[email protected]>
For the scan retry, previously we were comparing the entire vdi data
structure from the database using the (<>) operator. This is a bit
wasteful and not very stable. Instead let us just compare the vdi refs,
since the race here comes from `VDI.db_{introduce,forget}`, which would
only add/remove vdis from the db, but not change its actual data
structure.
Also add a bit more logging when retrying, since this should not happen
very often.
Signed-off-by: Vincent Liu <[email protected]>
When add leaked datapath: 1. add leaked datapath to Sr.vdis 2. write to db file 3. log enhance If there are storage exceptions raised when destroying datapath, the procedure fails and the state of VDI becomes incorrect, which leads to various abnormalresults in subsequent operations. To handle this, the leaked datapath is designed to redestroy the datapath and refresh the state before next storage operation via function remove_datapaths_andthen_nolock. But this mechanism doesn't take effect in current code. This commit is to fix this bug. Leaked datapath should be added to Sr.vdis to make the leaked datapath really work. And write to db file to avoid losing the leaked datapath if xapi restarts. Signed-off-by: Changlei Li <[email protected]>
There was some issue in the current code, from structure corruption to thread safety. Add some test and fix discovered issues. More details on commit messages.
When add leaked datapath: 1. add leaked datapath to Sr.vdis 2. write to db file 3. log enhance If there are storage exceptions raised when destroying datapath, the procedure fails and the state of VDI becomes incorrect, which leads to various abnormalresults in subsequent operations. To handle this, the leaked datapath is designed to redestroy the datapath and refresh the state before next storage operation via function remove_datapaths_andthen_nolock. But this mechanism doesn't take effect in current code. This commit is to fix this bug. leaked datapath should be added to Sr.vdis to make the leaked datapath really work. And write to db file to avoid losing the leaked datapath if xapi restarts.
For the scan retry, previously we were comparing the entire vdi data
structure from the database using the (<>) operator. This is a bit
wasteful and not very stable. Instead let us just compare the vdi refs,
since the race here comes from `VDI.db_{introduce,forget}`, which would
only add/remove vdis from the db, but not change its actual data
structure.
Also add a bit more logging when retrying, since this should not happen
very often.
Signed-off-by: Edwin Török <[email protected]>
GabrielBuica
approved these changes
Dec 12, 2024
psafont
approved these changes
Dec 12, 2024
robhoes
approved these changes
Dec 12, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Can't solve the conflicts on feature/perf due to branch restrictions, have to open PR from another branch.