Still, freezing is depending on the crypt-type of ECM's your oscam processes.Originally Posted by mikie8
If you have SKY UK ecm's (NDS)in cache, 6seconds is way too high.
For instance, 3 hops away someone is requesting an ECM for SKY.
If your oldest ecm in cache is 6000ms old, at time the hop3 peer receives it, including latency and time-out settings at other hop's, the cached ECM is mayb. 8000ms old, which is way to high for NDS. (and others).
I think there is no need to recompile whole oscam only say to your peers that is needed migration to cccam.
Or you met that whole module camd35 without peers invoke high mem consummation ?
Correct.. Therefore i disabled it in the whole build.
But.. After one hour running smooth withoud camd35... Mem is rising again with 1MB/10 seconds.
So.. I start doubting on the cause again... :(
Disabling cache-ex in the build DOES solve the problem, 100% sure..
But i thought (-en was hoping) that it was the combination of CAMD35 and CACHE-EX together..
Try to change account / readers with setting like this
http://www.satsupreme.com/showthread...l=1#post928095
Because in past i make settings like this and cost me high load
USER:
group: 1 or 2 or 3... (do not leave blank !!!)
cacheex mode: 3 (reverse cache push)
cacheex max hops: 10
Anticascading numusers: 0
ccc maxhops: -1
ccc reshare: 3
ccc ignore reshare: 1 use reshare level of reader or user
ccc stealt: enable: use oscam <> cccam protocol
keepalive: ON
READER:
group: 1 (or all groups you are using) > so can aso be 1,2,3,4,5 etc > remember that the group you use in the user settings must be also the same here ! so if in user group = 10 then reader must have at
least group 10 !
load balance: 100
caid: < LEAVE EMPTY !
drop cw's with wrong checksum: NO
cacheex mode: 3 (reverse cache push)
au disabled: yes
cache ex maxhop: 10
inactivity timeout: 30
reconnect timeout: 2
request timeout: 0
version: 2.3.0
maxhop: 0 < these settings are needed this way, otherwise your other clients will see the cache as cards !
mindown: 15 < these settings are needed this way, otherwise your other clients will see the cache as cards !
reshare: 3
want emu: NO
keep alive: NO
CONFIG:
extra delay: 500
client timeout: 6000 < IMPORTANT
fallback timeout: 1600
cache delay: 10 < IMPORTANT
client max idle: 120
global sleep: 0
suppress cms08: NO
reader restart seconds: 5
drop duplicate users: NO
max cache time: 8 < IMPORTANT
max cache count: 2500
ecm doublecheck: NO
cacheex wait time: 50
cacheex write statistic: OFF < IF YOU PUT THIS ON, CPU WILL GET A HARD TIME.. ONLY SET TO ON IF YOU WANT TO DEBUG !
Hello,
A few days earlier, I changed my protocol from cccam to camd35. Server behaves exactly the same. I have about 25 normal readers and 7 cacheex mode: 3. The version of oscam is svn7212.
Consumption of CPU resources and mem, was the same in both protocols.
Altough I honestly appreciate your help and suggestions, but i'm not gonna change 60 readers into these suggested values..
I specifically do not agree with:
Cache-ex maxhop: (i don't want to send/receive cache from hop 10, i want to send/receive cache from hop 2/3 max)
maxhop/mindown/reshare: (parameters only applies to normal readers, not cache-ex readers)
Also the "Config" part contains values that i can't understand as stated above..
But i just set them according to your suggestion, and doesn't make a difference to the mem usage, so i reverted back to my own.
this is wrong.
READER:
group: 1 (or all groups you are using) > so can also be 1,2,3,4,5 etc > remember that the group you use in the user settings must be also the same here ! so if in user group = 10 then reader must have at
least group 10 !
this will cause a loop ,if both peers use the same settings, 1 will send to 2 who will send to 1 who will send to 2 untill it timeouts .
reader and user settings for the same peer must not include the same groups
try disabling block ip/names in builds after 7210 ,i suspect it may be the problem .
im sticking with a build lower than 7210 for now and have no problems with cache ex and camd35
Hmm...
Is it correct that these two parameters (blocksameip / blocksamename) cannot be set from the config file ??
I can put both to "NO" during runtime, but after reboot it reverts to "YES", nothing is stored in oscam.conf
Adding the parameters manually to the config (=0) doesn't work either..
Having both to "NO" during runtime doesn't solve my issue, mem is still rising rapidly
I know that is not right configs i mention that there is extra unneeded setting that cost me a load (my groups settings are right) according to this.
User Group 1 - Reader Group 2,3,4
User Group 2 - Reader Group 1,3,4
User Group 3 - Reader Group 1,2,4
User Group 4 - Reader Group 1,2,3
Here is some tricks from me, to reduce cacheex=3 memory consumption for both camd35 and mostly for cccam like cacheex server.
In account section set:
suppresscmd08 = 1
cccmaxhops = 0
cccreshare = 0
In reader section set:
For camd35:
ccchop = 1
For cccam:
cccmaxhops = 0
cccreshare = 0
cccmindown = 0
Posting Permissions
