Cache-EX Loopback

[galiano]

I use svn version 8538 with no issues - i don't know why folk keep using latest versions? if it ant broke then don't fix it!

That you are using an older version of oscam, it does not matter, because more than 90% of cacheex peers used the latest versions. Which means that the flow cache that you receive is complete with fake cwc. It is true that there is something wrong with cachex, but the older version is no solution.

Yesterday in streamboard is open new topic related with cacheex. As I think, most interesting question in this topic is still without answer:

I also do not understand the reason for the huge cache flow in EX_3 mode. One thing is certain: most of it is unusable cache, too old, too slow, or some other reason (in other words, fake)

An example:
If using caceheex more3 for one provider, for example 1702, which has approximately 150 sids, which should mean that the cache size 150 contains cwc to all 150 sids. Yes, but it actually is not. All we have seen even have a cache size 5000, it still you can not get 100% hits.
Another question is: why does cache to one provider with 150 sids, size reaches 5000, in theory, all duplicate (identical) cwc, should be deleted?

[CapNCooK]

That you are using an older version of oscam, it does not matter, because more than 90% of cacheex peers used the latest versions. Which means that the flow cache that you receive is complete with fake cwc. It is true that there is something wrong with cachex, but the older version is no solution.

Oscam is not the source of the bad CW. Put two oscams with local cards next to each other and let them generate and exchange cache.
Total size will match the the size expected, without any problems.

So, before we link "older and/or newer versions" to "bad cw", there is no reason!

Bad CW comes from external sources, CSP, MultiCS, self-coded cache banks, cards in penalty mode, people that inject bad cache to disturb scene, etc.

--

About duplicates.. (ECM header with different CW in tables) Yes, it is deduplicated.. And Yes, each ecmhash+CW in table is unique.
Question remains.. Where does all this come from.. I have no idea either.. But it IS stored as received.

--

Yes, there is much to improve in SVN's (new and older), have some patience.. It will be rewarded...

[galiano]

Viz. That's what I meant. All problems with the huge cache size began when was added the ability to share via CSP and Multics. This brought more trouble than how good.

[turbopower]

I've analyze cacheex 3 function from multics r72 and I think there is no thread because is based on oscam logic. But for direct cache from csp and multics you are right. But let's be focused about one thing multics and csp are very different but have ability to exchange cache each other without any dead cw and other issues but oscam is not capable to manipulate the same cache. This mean that the problem is inside into oscam code not in csp and multics. So for me in this moment oscam is amazing and only solution to read cards but for cache and multi peers regular exchange is total disaster.

[CapNCooK]

.. but oscam is not capable to manipulate the same cache. This mean that the problem is inside into oscam code not in csp and multics.

Sorry, but that on itself is just non-sense. Point to some proof along that statement. (so it can be fixed), IMHO oscam is not broken.

[mikie8]

i get 80% cache hits , glitch free viewing . oscam does all i ever want from it .

[turbopower]

Sorry, but that on itself is just non-sense. Point to some proof along that statement. (so it can be fixed), IMHO oscam is not broken.

I have only one prove - if you try to use multi/csp cache with multics for example you will see that everything is fine (working normally). This mean itself that the cache is correct but oscam can't manipulate with them. So solutions are 2 - oscam team will fix the issues or csp/multics cache bridge should be disabled in oscam completely.

[CapNCooK]

And how do you exchange between Oscam and MultiCS ?
And what does not work exactly ?

CSP => Oscam(UDP) works.
MCS => Oscam(UDP) works too.

[turbopower]

And how do you exchange between Oscam and MultiCS ?
And what does not work exactly ?

CSP => Oscam(UDP) works.
MCS => Oscam(UDP) works too.

I mention in general about fake cache that coming from CSP/Multi. Because the cache itself is not fake but oscam don't have successful mechanism to control it and only solution is to disable it from oscam config files.

[CapNCooK]

I mention in general about fake cache that coming from CSP/Multi. Because the cache itself is not fake but oscam don't have successful mechanism to control it and only solution is to disable it from oscam config files.

MCS, is using same methods for cachecreation and storage as CSP. Thing that gets stored is a Hashed ECM request (also called CSP Hash) + CW and some metadata, and oscam processes that very well. So.. The problem is where exactly ?