Cluster session concurrent write

[Selvatico]

Hello to all,

So problem is next - we have several workers in cluster. 2 of them receive requests almost at the same time - each of them read session storage (MongoDB in our case). In parallel they are finishing their work and save all values of from session

request.session._store

to cache storage. And one of them override value of another, because both of them doesn't see changes of another and dump all data to storage. I'm wondering if exist some technics to avoid such behavior using yar ? Because from my point of view - we should implement in yar some atomic updates, but it should be implemented in catbox adapters too, which requires too much changes in core of Hapi.

[PavelPolyakov]

@OmniJeff
Hi, what is your view on the parallel writes?
How this should be done according to the architecture of the application ? Or how could we avoid this?

[OmniJeff]

This would be a problem for catbox, which is what yar uses as its persistence layer. yar has little control over how catbox caches data.

[OmniJeff]

Will not change catbox behavior. Good explanation here: outmoded/catbox-mongodb#13 (comment)

[lock]

This thread has been automatically locked due to inactivity. Please open a new issue for related bugs or questions following the new issue template instructions.