Page MenuHomePhabricator

Hook in to status of HAProxy in front of TTS
Closed, ResolvedPublic

Description

Speech synthesis is the part that risk being overloaded.

For non horizontally automatically scaled environments we need to limit user requests on high load, or at least let the user know there is a queue and how long it might be in seconds. We currently handle this by adding a request queue using HAProxy. It's possible to extract status via https://www.haproxy.com/blog/exploring-the-haproxy-stats-page/, so we should probably create a request to that prior to sending the request to Speechoid server.

However, this also means that we need to be aware of what TTS what language is using.

Event Timeline

Change 695273 had a related patch set uploaded (by Karl Wettin (WMSE); author: Karl Wettin (WMSE)):

[mediawiki/extensions/Wikispeech@master] Use connection queue against Speechoid TTS operations

https://gerrit.wikimedia.org/r/695273

Change 695273 merged by jenkins-bot:

[mediawiki/extensions/Wikispeech@master] Use connection queue against Speechoid TTS operations

https://gerrit.wikimedia.org/r/695273