You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When running as a bot user account (with ratelimits), some actions (such as mass-redactions and unban re-inviting) may fail as the current ratelimit handler does not respect the ratelimit response. As a result, the ratelimiter has a low chance of success, as repeated ratelimited requests to Synapse will increase the ratelimit exponentially, faster than Draupnir's fallback does.
The text was updated successfully, but these errors were encountered:
The first issue is about to be fixed in an upcoming pull request, where the rate-limit response is incorrectly handled due to a check for an always undefined property in the response error object.
The second issue is that rate-limits are difficult to predict and categorise, meaning that when a rate-limit is encountered, it has to be waited out before additional actions can be taken (for example, if a redact batch succeeds in rooms A, B, and C, but gets rate-limited in D, rooms E, F, and G have to wait until D's rate-limit expires, even if they would not have been rate-limited themselves)
* Fix the individual request rate-limit
This fixes the first problem raised in #693
---------
Signed-off-by: nexy7574 <[email protected]>
Co-authored-by: Gnuxie <[email protected]>
When running as a bot user account (with ratelimits), some actions (such as mass-redactions and unban re-inviting) may fail as the current ratelimit handler does not respect the ratelimit response. As a result, the ratelimiter has a low chance of success, as repeated ratelimited requests to Synapse will increase the ratelimit exponentially, faster than Draupnir's fallback does.
The text was updated successfully, but these errors were encountered: