Clear cache from empty/zero ECM

[mumbua]

I have a Server with over 100 Peers. After deaktivating CSP the Hits comes higher. The Traffic and Cpu load goes extremly down.
Is better to set drop csp Cache and cacheex max hop to default. Default is 10.
My Friends Server has only 30 Peers. I set also drop csp Cache. The hits get less -12 %.
After i activate on one User CSP and leave it at max hop 10. I get the loost 12 % back. but Traffic double. CPU Load 3x times higher.
Also is better to look for more Peers with Cacheex Cache and all to max hop 10.

[nikko]

I do not think that hop10 it is very appropriate for NDS caids.This will generate only old CW.

[CapNCooK]

I do not think that hop10 it is very appropriate for NDS caids.This will generate only old CW.

Why ?

Cache is there.. Or not.. Independent to which hop it came from..

In an ideal situation you would have 10 x 10ms latency.. so cache flows in @ about 100ms..

--

Something else.. I dont know for sure if oscam can use CSP cache or not..
But i do know that turning it on/off on all peers has a big effect on my 0963 hits..

The rest seems to hit as normal.. But i can reproduce a 15% ~ 25% drop without having the CSP cache.

--

To know for sure, i posted this question on streamboard, but get no answer.. (yet)

I am starting to believe that oscam indeed can not use the CSP cache, so i rather suspect the 'drop CSP Cache'
code to drop too much cache for some reason.

[m404]

Cache is there.. Or not.. Independent to which hop it came from..
In an ideal situation you would have 10 x 10ms latency.. so cache flows in @ about 100ms..

this ... so many times this.

i believe ALOT of cache exchange people do not understand that in a from VPS to VPS (or any other hosted server to hosted server) scenario, the ping times are usually VERY low.
i can ping, from my server, to most of my peers, with pings below 20ms ... who cares if a single ECM reply is now hopped 10 times with those latencies?
yes, i understand, the workload a server does on a package takes some time too, but nontheless, the timings are STILL more than just acceptable.
the reason in CCcam and other protocols we used to hate hops is because the system was completely different ... it was based on a handshake-request-search-reply-handshake system, which cause ALOT of latency ... cache is being pushed (in mode 3, which we almost all use) whether you request it or not ... the timings are completely different, and as CapNCook correctly adresses ... if the cache is there, and can be used, then who cares what it'll show as a timing?

[CapNCooK]

And to put some more sense in this thread..

I just spoke to the developer that has created the "drop CSP cache" feature.

--

1) Oscam CAN use CSP cache, because it has alternative methods to recognize from CSP cache

2) Currenty, In all CSP Cache the PROVID (eg. 00006A) gets lost. This is a problem, that we look into now with the developer.
It might require a CSP codechange to get this corrected.

3) Cache with (default) PROVID 000000, eg: 0963:000000 DOES get recognized. So blocking CSP cache will lower hits on these CAID's.

--

It would be nice to see people stating from knowledge.. not from assumptions.

[nikko]

Real facts not assumptions friend.
More than 30% of the cache are empty CW.
If using filter empty CW gone and CPU usage decreases.
Everyone can see it for yourself with debug -d512.

[CapNCooK]

Real facts not assumptions friend.
More than 30% of the cache are empty CW.
If using filter empty CW gone and CPU usage decreases.
Everyone can see it for yourself with debug -d512.

Yes. Offcourse load will go down when you filter away 50% of your cache stream.
No-one is doubting that.

You were stating:
- that oscam cannot use CSP cache.
- cacheex_maxhop=10 cannot be used

And.. you cannot see the content of cache thru -d512.
You only see hash.. and when configured logfrmt properly.. mayb checksum too.

Not all logged cache with 000000000 is "empty".., its just "CSP".

[nikko]

I've never said that oscam cannot use CSP cache.
I personally share with CSP users. With enabled filter still receive cache of them.
About hop 10 - 10x10ms = 100ms latency if forget about server response time. But in reality many users have poor adsl, cheap routers, wirless links and use internet bandwidth for other purposes-torrents,streaming video,IPTV....
Аnother question is what hardware will be needed to receive cache to hop10 from 100+ peers.
This i see with -d512 log without filter.
22:48:08 5DF4D8 c got pushed ECM 0963&000000/0000/156A/00:00000000000000000000
000000000000 from cache
22:48:08 5DF4D8 c got pushed ECM 09CD&000000/0000/2B0F/00:00000000000000000000
000000000000 from cache
what this means?

[CapNCooK]

I've never said that oscam cannot use CSP cache.
I personally share with CSP users. With enabled filter still receive cache of them.

You advise people to turn off CSP cache. In order to drop "empty CW".

This option not affect the usable CW only empty.

Which is not true.

About hop 10 - 10x10ms = 100ms latency if forget about server response time. But in reality many users have poor adsl, cheap routers, wirless links and use internet bandwidth for other purposes-torrents,streaming video,IPTV....

You know that on average the validity of a CW is about 10 seconds ? And some even longer ?
So even if the cache has 4 seconds latency (which it will never reach) .. it still can open up the channel for these oscam's that zap onto it.

What about irdeto cache. this may travel quite long. Seca too.

--

Oscam will never use "wrong" or "outdated" cache. Cache is just there.. Or not.

--

Аnother question is what hardware will be needed to receive cache to hop10 from 100+ peers.

Big hardware (i7, xeon), and a good line (20Mbit up and down)

This i see with -d512 log without filter.
22:48:08 5DF4D8 c got pushed ECM 0963&000000/0000/156A/00:00000000000000000000
000000000000 from cache
22:48:08 5DF4D8 c got pushed ECM 09CD&000000/0000/2B0F/00:00000000000000000000
000000000000 from cache
what this means?

See here: http://www.streamboard.tv/wiki/OSCam...am.conf#ecmfmt

[superg1972]

Hi all,

i have read this post 10 times but i still cannot understand what the "drop CSP cache" feature does.


CapNCook wrote:

1) Oscam CAN use CSP cache, because it has alternative methods to recognize from CSP cache

3) Cache with (default) PROVID 000000, eg: 0963:000000 DOES get recognized. So blocking CSP cache will lower hits on these CAID's.

So from this i can understand that the "drop CSP cache" feature will drop ALL the cache that is coming from CSP, good and bad (eventually).
Is it correct my assumption?

Thank you