cookie time length v LSWS cache lifespan when using VARY

[mkaaaay]

We use a VARY statement (based on user being logged in or not) in our application of LSWS caching, and sometimes find weird anomalies. I'm thinking it maybe something to do with the timing of LSWS cache clearing and the lifespan of the logged in cookie.

Given a cookie start time is almost never going to coincide with the start time of he LSWS cache my questions is this:

Q: should they equal each other? or should one be a bit longer than the other perhaps, so there is some kind of overlap or sorts?

Thank you.

 

[serpent_driver]

should they equal each other?

This question implies you use private cache for logged in users? If yes, why, if you have "punch holes" defined for critical (private) content like shopping cart?

Like in almost every application normally there is no need to bind TTL of logged in cookie to LSWS cache TTL.

 

[mkaaaay]

no private cache, public varied cache for logged in users based on a cookie existing.

 

[serpent_driver]

Okay, but why bind cookie life time to cache life time if cache is public? Sorry, this construction confused me. In my mind cookie TTL for logged in users should only have relation to session life time, but not cache TTL. Maybe you should give us more details why you ask for such settings.