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
Investigate startup time regression #380
Labels
blocks release
used in exceptional circumstances when the `latest` tag would break
L4 Most Users
Likelihood
needs investigation
P4 Blocks All Development
Priority level - breaking build / dependency / test
T5 Major usability
Impairs usability in key scenarios.
Comments
Gnuxie
added
L4 Most Users
Likelihood
P4 Blocks All Development
Priority level - breaking build / dependency / test
T5 Major usability
Impairs usability in key scenarios.
needs investigation
blocks release
used in exceptional circumstances when the `latest` tag would break
labels
May 4, 2024
Gnuxie
added a commit
to Gnuxie/matrix-protection-suite
that referenced
this issue
May 4, 2024
We believe this is the cause of a performance regression in Draupnir: the-draupnir-project/Draupnir#380. Unfortunately we don't have the infrastructure in our unit testing or deployment to be able to truthfully understand the cause. So we have created a follow up the-draupnir-project/Draupnir#389. Our approach for now is just to guess some of the most glaring changes that have caused the regression. If this fails to make a significant difference, other culprits are changes to the parsing code (ie, there was a recent version of MPS where we moved away from accidentally using the empty schema for most event content), including the recent upgrade to typebox 0.32.22.
Gnuxie
added a commit
that referenced
this issue
May 4, 2024
This contains a workaround for this issue #380 please see Gnuxie/matrix-protection-suite@e57f266.
Merged
Gnuxie
added a commit
that referenced
this issue
May 4, 2024
This contains a workaround for this issue #380 please see Gnuxie/matrix-protection-suite@e57f266.
Ok workaround didn't work. |
Gnuxie
changed the title
Investigate why the backing store takes a huge penalty from concurrent calls when it is supposedly synchronous
Investigate startup time regression
May 5, 2024
Gnuxie
added a commit
to Gnuxie/matrix-protection-suite
that referenced
this issue
May 6, 2024
Gnuxie
added a commit
to Gnuxie/matrix-protection-suite
that referenced
this issue
May 6, 2024
sinclairzx81/typebox#825 (comment) the-draupnir-project/Draupnir#380 Co-authored-by: Haydn Paterson (sinclair) <haydn.developer@gmail.com>
Gnuxie
added a commit
to Gnuxie/matrix-protection-suite
that referenced
this issue
May 6, 2024
Gnuxie
added a commit
to Gnuxie/matrix-protection-suite
that referenced
this issue
May 6, 2024
sinclairzx81/typebox#825 (comment) the-draupnir-project/Draupnir#380 Co-authored-by: Haydn Paterson (sinclair) <haydn.developer@gmail.com>
Gnuxie
added a commit
that referenced
this issue
May 6, 2024
Part of #380. We've updated typebox and changed the way events are decoded.
Merged
Gnuxie
added a commit
that referenced
this issue
May 6, 2024
Part of #380. We've updated typebox and changed the way events are decoded.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
blocks release
used in exceptional circumstances when the `latest` tag would break
L4 Most Users
Likelihood
needs investigation
P4 Blocks All Development
Priority level - breaking build / dependency / test
T5 Major usability
Impairs usability in key scenarios.
At startup we changed the code in MPS to load rooms from the backing store all at once rather than one by one
https://github.com/Gnuxie/matrix-protection-suite/blob/6ae7c8ce8226cb18eea941345d9907d62aa75b5d/src/Protection/ProtectedRoomsManager/StandardProtectedRoomsManager.ts#L103
https://github.com/Gnuxie/matrix-protection-suite/blob/c3e574a1f7740e5c8057fbf43e9c98d9bd32e640/src/StateTracking/StandardSetRoomState.ts#L50-L52
-- or so i thought, but they should in effect be exactly the same? Either way on breadpirates.chat i noticed startup going from a 10seconds to ~1minute after this change, which is about as long as it takes without the backing store... so not sure what is going on.
The difference between
main
andv2.0.0-beta.2
looks like this:I suspect that the loading code is a red herring, and in reality the changes are due to changes to typebox and the schema in the same release of MPS
The text was updated successfully, but these errors were encountered: