if using a build after 7210 it will be in config in the webui .Originally Posted by CapNCooK
if using a build after 7210 it will be in config in the webui .
it would seem its a config conflict , as i get memory fill with cccam and camd cache ex mode 3 connections . not all just one cccam and one camd peer .
other camd and cccam peers work fine with no mem fill.
Decrease Cache-EX Maxhop: with one point bellow in account and reader section. This will help you to fine tuning the load.
In my case value of account is 4 in reader is 5 (to prevent loosing 1 hop down to my peers).
the cause is ring transmissions , peers are sharing with same peers causing loops 2 hops away .
need to contact cacheEx3 peers and see if you share peers and then disable group for shared peer .
Yes this the big problem in cacheex exchange. For this reason I want from most of my peers to send me his config about our exchange for checking.
Well, as hopefull as that sounds.. I'm still not entirely convinced..
In my case, if i disable all my cache readers, and enable all my cache users, mem is still filled up.
So any form of "looping" is eliminated as cause by that.
Disabling some cache users slows the memfill, but its still rising.
Disabling all cache users stops the memfill completely. (even with all cache readers enabled).
Also i tried dozens of builds last night, from 7050 to 7229, all show the same behaviour.
disabling wont help , it confused me for a while , as i thought like you , its logical to assume if the traffic is one way no loops are there . but assumptions are just fallacies by another name .
put all your cache readers ( cx 1 & 2 ) or users (cx 3 ) in different groups from each other , then in users ( cx 1 &2 ) or readers (cx 3 ) put only groups with no cacheeEX .
you will then receive cache to your cache3 and that cache3 will be shared to others cache , no cache will be sent directly to another cache bypassing your cache3 . you become a buffer .
you should then be able to enable all readers and users without mem fill .
then start to share cache directly between cache peers,till you get mem fill . you should find who is looping this way .
i have 3 peers i will be contacting as if i share a group of one to any other i get mem fill , i can share them with others no problem ,just not amongst them 3 .
this is why i suspect ring transmissions a hop away , oscam watches the caches progress and as it hops in a loop oscam fills mem with cache reports or something . i dont read code so what exactly is happening is not known by me ,this is just from testing .
Hmm. Thx mikie8, i will investigate further.
For anyone that uses Oscam Cache-EX with > 10 cache-ex peers
=> Can you please report your current build and linux OS ? (incl. 64 bit or 32bit)
=> Also report if you have issues or not
Thanks!
Oscam 1.20-unstable_svn build #7059
Debian 2.6.32-5-686 #1 SMP Thu Nov 3 04:23:54 UTC 2011 i686 GNU/Linux
Issues of course with processor/memory utilisation
but things getting stabilized with this setting:
in oscam global config:
nice = 20
in debian server console
ulimit -m 100000
ulimit -v 100000
And after putting ulimit, restart oscam
last 30 min oscam running pretty well, no oscillations at all
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
11982 root 39 19 58212 15m 1696 S 5 0.5 1:45.60 oscam
Ring transmissions mentioned above is not big deal for cacheex mode 3 because CX3 actively ignore/reject duplicate ecms, CX2 mode may have some issues which can result into black screen sometime.
And dont forget that cache is global and shared between all, no metter which group you use
Cheers
Hello htcX,
i tried the ulimit cmds as suggested, but after Oscam doesnt pass the init state, and does not start.
In fact, it tries to restart about 5 times / second from seeing the same inits pass in the oscam.log.
As i'm not familiar with the command, i don't know how to alter it in the correct way...
what do you suggest htcX ?
Thx!
Posting Permissions
