Replies: 5 comments 3 replies
-
|
Agreed, I suspect there are a lot of workflows that could be reduced in frequency or made more efficient. It would be interesting to see which projects used the most resources. |
Beta Was this translation helpful? Give feedback.
-
|
I agree it's good to give projects insight into the resources they use, and motivate them to reduce that (within reason). As you mention https://infra-reports.apache.org/#ghactions already exposes this information for the projects you're part of (AFAIK). I kind of fear that if we'd provide this information across all projects, any project that's not in the top-5 will look at it and go "ah no problem then", which I'm not sure is what we want... |
Beta Was this translation helpful? Give feedback.
-
|
Rather than rank the projects in strict order, maybe it would be possible to rank them in general terms. These categories would need to be refined as time goes on. === As to the infra-reports data, at present I don't find it useful. |
Beta Was this translation helpful? Give feedback.
-
|
thanks for your comments @sebbASF and @raboof I'd like to have a go at estimating the emissions . Do you know who would be able to access the reports from the URL above to get the stats at org level? Would save me having to write a scraper. Would be interesting to compare with the figures from eco-ci There are 2 tracks from there:
|
Beta Was this translation helpful? Give feedback.
-
|
I opened a ticket for infra to get access to the global stats A quick estimate gave me the following figures, just to put things in perspective. Apache Tika - 30 days of GH actions now Apache Superset, which I think has the top number of stars on GH A quick breakdown of the coefficients:
There would be more things to add to it like PUE etc... and compare with other sources but this is just to give a rough estimate. 20kg a month (i.e eq drive 160km in an average petrol car) probably does not seem that much but over a whole year and across the whole ASF estate, we would be getting a pretty substantial impact. Yes, these minutes do not cost anything to the ASF (being in public repos and on standard runners) but they have an environmental impact. The good news is that we can do something about it. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
The stats about GitHub actions can be obtained per repository but given that the ASF has nearly 3K of them, aggregating them is not very straightforward.
The same stats can be obtained at the org level, e.g. for the last 30 days. It is possible to download the report as a csv
Why these stats? I think it would be great to be able to estimate the environmental impact of running GH workflows. The default runners provided by GH are located in the US, where the electricity grid is relatively carbon intensive compared to other locations due to the energy mix (see ElectricityMaps for a live view).
Longer term this could be improved by having ASF supported runners in cleaner locations like Canada or Sweden, like https://carbonrunner.io do.
Estimating the current emissions as a baseline would be a great initial step. I'd be happy to help with the data and methodology but need the stats across the ASF.
It would be good to make this information available to individual projects to raise awareness (and as a motivation to reduce usage in general), maybe as part of [https://infra-reports.apache.org/#ghactions].
Beta Was this translation helpful? Give feedback.
All reactions