elfkuzco
commented
5 months ago Taking into account the location of users, location of mirrors, individual content of the mirrors, capabilities of each mirror (how much load they can take) and the online status of the mirrors, the logic for the selection of a particular mirror is done with the following considerations:
- Using the GeoIP Database to look up the client's IP address, MirrorBrain can choose a mirror from the same country. If no mirror in the same country is available, one from the same continent is picked. In addition, it is possible to define fallback mirrors that are known to be suitable for a certain country, thereby optimizing over a random choice. Those fallback mirrors are used only in reserve, when a mirror in the country itself does not exist.
- In addition, MirrorBrain can use an Apache module named mod_asn to find out more about the client and to exploit network locality. By looking up the clients network prefix and autonomous system, MirrorBrain can match these to the mirrors. Thus, a request from the mirror's own network will go to exactly that mirror, and the same is true for the autonomous system the mirror is in.
- According to this issue, geographical distance to the mirrors is considered using a computation which involves a simple approximation of the Haversine formula to calculate the closeness of mirrors to the client.
distance = sqrt((x2 - x1)^2 + (y2 - y1)^2)This helps for countries that have no mirror themselves, where only a random mirror from the continent could be selected otherwise.
- In cases of regions that don't have mirrors, and where a geographical choice would lead to insufficient results, fallback mirrors can be defined that are known to handle that region well.
- Individual mirrors can be configured to serve only requests from their continent, their country, their autonomous system, or their network prefix. This is important for mirrors with limited bandwidth, or for countries with poor internet connectivity.
According to a section of the video conference, MirrorBrain works in the background. As such, there is no direct way to override the selection decision. However, collecting download statistics can help the content creators define fallback mirrors for countries with low internet connectivity as there is a bit of randomization done even after taking into account the closeness and strength of a mirror. This is highlighted in this section of the video conference where the closest mirror might not always be the best.
kelson42
rgaudin
In order to limit the scope of this project, and to understand what are the available variables we can act upon, we need to have a very clear understanding of how the load balancer works: what information is used and how in order to compute a request response.
I suppose it's all in mirrorbrain's documentation.
I am sharing here the configuration (removed the admin's details) as some would matter.
mirrobrain.csv
@kelson42, from our recent call, I suppose the
*_onlyparams might not be properly set.