Change Details

Note: this is a draft task. It should be reviewed by @nnikkhoui before it is tagged and assigned. == Goal Get experience on Toolforge capabilities and its suitability for use in hosting future PoC services == Acceptance Criteria - [ ] Determine CPU constraints and associated impact on foreseeable service needs, such as data processing and query execution. - [ ] Determine memory constraints and associated impact on foreseeable service needs, such as data processing and query execution - [ ] Document all findings on-wiki The findings from {T280849} and {T280851} may be helpful for this task. The results of this analysis should be compared and contrasted with the results from {T280850}

Note: this is a draft task. It should be reviewed by @nnikkhoui before it is tagged and assigned. == Goal Evaluate Toolforge capabilities and their impact on hosting the image suggestions api service. Consider implications for hosting future PoC services == Acceptance Criteria - [ ] Determine CPU constraints and associated impact on the image suggestions api service needs - [ ] Determine memory constraints and associated impact on the image suggestions api service needs - [ ] Extrapolate findings to provide guidance for future possible services - [ ] Document all findings on-wiki The findings from {T280849} and {T280851} may be helpful for this task. The results of this analysis should be compared and contrasted with the results from {T280850} The Toolforge documentation spells out certain limitations and constraints: [[https://wikitech.wikimedia.org/wiki/Help:Toolforge/Kubernetes#Quotas_and_Resources | documentation ]] This task is about contextualizing that information. Regarding the image suggestion api, what operations have proven to be most affected by these limitations? Are there any reasonable mitigations or workarounds that could be implemented? Regarding future services, what questions should implementors of future services consider when deciding whether or not to use Toolforge? What characteristics of future services might make Toolforge an appropriate or inappropriate choice?

Note: this is a draft task. It should be reviewed by @nnikkhoui before it is tagged and assigned. == Goal Get experience onEvaluate Toolforge capabilities and its suitability for use intheir impact on hosting the image suggestions api service. Consider implications for hosting future PoC services == Acceptance Criteria - [ ] Determine CPU constraints and associated impact on foreseeable service needs, such as data processing and query execution.the image suggestions api service needs - [ ] Determine memory constraints and associated impact on foreseeablethe image suggestions api service needs, such as data - [ ] Extrapolate findings to processing and query executionvide guidance for future possible services - [ ] Document all findings on-wiki The findings from {T280849} and {T280851} may be helpful for this task. The results of this analysis should be compared and contrasted with the results from {T280850} The Toolforge documentation spells out certain limitations and constraints: [[https://wikitech.wikimedia.org/wiki/Help:Toolforge/Kubernetes#Quotas_and_Resources | documentation ]] This task is about contextualizing that information. Regarding the image suggestion api, what operations have proven to be most affected by these limitations? Are there any reasonable mitigations or workarounds that could be implemented? Regarding future services, what questions should implementors of future services consider when deciding whether or not to use Toolforge? What characteristics of future services might make Toolforge an appropriate or inappropriate choice?