Cache-EX Loopback

[galiano]

To all intelligent people, it is clear that users of MCS and CSP, used many tricks with purpose, their size cache to look better (these are two proven fakers software).
Here the question arises: why was it necessary, oscam team, add to its good working code, communication with MCS and CSP, with no ready solution to address their shortcomings? I am convinced that oscam works much better without communicating with these emulators. Who wants to use MCS and CSP, can to install, their clean versions do not have to go through oscam. Incidentally, the current state of oscam, and specifically its cacheex function, causing a huge load, with small hits, prompting more and more people switch to the above emulators (in this topic, I see several such).

Comparison:
Before being added, the ability to communicate with MCS / CSP, my size cache, was 1000-1200, so I was getting over 90% hits. Now I also get 90% hits, BUT now my size cache 10000 and my PC is five times more powerful. That is ten times larger size cache, and the final result (hits%) is the same. Mathematics is an exact science and the facts are undeniable. I would not say that this is a positive development.

[turbopower]

I mean exactly what you says galiano. Thank you my friend. In now situation oscam is very tricky for oscam cache exchange if you have lot of peers. I'm not able to spend more than 50 euro for hosting but I want to have exchange with all of my friends. So oscam is useless for me right now. I think CSP cache connectivity must be removed from oscam for better.

[CapNCooK]

I consider myself intelligent, so i guess i am eliable to answer.

If i push cache from MCS to Oscam, it hits in oscam
If i push cache from CSP to Oscam, it hits in oscam
If i push cache from Oscam to CSP, it hits in CSP
If i push cache from Oscam to MCS, it hits in MCS

In all scenario's, correct cache-size is reported in all mentioned software.

So what EXACTLY does not work ? I asked that a couple of times now, but.. no answer!
No clear answers => No solution

The only thing i read that makes sense, is that some "fake CW" generator exist for CSP/MCS and its beeing used.
As soon someone write a simular "fakers" software for Oscam, and injects at some point, the same problem occurs there.
Removing ECM_Handler and/or removing CSP connectivity is NOT a "solution".

If we can just agree that without any external factors, like "fake CW simulators" and all that, cache-ex + MCS/CSP works fine, than this discussion is clear and solved.

Next topic could be, how to prevent bad cw in cache.. But not title it "oscam is broken", because it has really nothing to do with anything.

[bunbunsat]

I'm no expert in Cache-EX (even by far) so I don't know the in's and out's of the internal workings. From what I read and understand CSP (and I think MCS also) drop the PROVID. My best guess is this cache will prevent good deduplication.

e.g.:
ECM for 0100&00006A arrives in cache
Same ECM arrives from CSP or MCS but without PROVID. So the same ECM is in cache but for 0100&000000

Unless Oscam "recognizes" these as the same ECM you have 2 ECM's in cache which are basically the same. Because of this cache size can increase significantly without resulting in increased hits. To me that's a serious flaw. The flaw IMHO is not in OSCam but in CSP and MCS so they should fix this.

As a result I filter out 0100 extensively basically dropping 0100&000000. I might miss some cache but that does not weigh up to the fact that my cache-size is better under control.

[galiano]

I do not think the problem is entirely outside oscam.

An example:
Using build r8755 size cache is 2000-2500, hits 90-94%, after update to one of the latest builds (without changing the configuration and number of cache peers, same conditions) size cache, increasing more than five times, up to over 10 000 -12 000, hits% remains unchanged. Where does this traffic, as a real cache transver is equal to both oscam versions?

P.S.
@CapNCooK
I want to note that my comments are very well-intentioned and not directed personally to you! I perfectly know that you are doing a lot of amplifying the development of oscam, for which I am grateful. Despite some shortcomings, I think oscam is better than other cache-ex software.

Regards!

[bunbunsat]

You did not change your settings but who says one of your peers did not add more peers? Can you revert back to the old version. If your cache drops back down to its original size your assumption might be right. But this needs some testing.

[galiano]

@bunbunsat
I use a separate oscam build r8755 (reasons are described above) only for cacheex, for now I do not intend to switch to a newer version.

[ametovic]

thanks for that @galiano, I too was sitting on 12k in cache ..

back on r8755 and all is good... odd

[CapNCooK]

To ametovic and/or galiano, can you please check;

- Size change between 8755 and 8894
- Size change between 8755 and 8895

I want to make sure 8895 is causing the higher size.. If so, it is correct behaviour since max_count is removed in 8895.

[galiano]

Size increase much before 8894/5, before you remove the parameter max_count.

My advice to all users, using, separate oscam only cacheex is: use r8755, with a line contact with your oscam your mixed (mixed oscam, may be new, the latest version)
settings in r8755, should be:

max_time = 0
cacheexwaittime = 0
wait_time =

The above parameters are activated only in mixed oscam.


Thus size cache will decrease repeatedly without affecting the hits%. Accordingly, the CPU load will be reduced 3-5 times.
This is a little trick that you can avoid unnecessary high traffic.