best cach config require

[_pinky_]

Till oscam 9121 i use cacheex 2 with protocol = cs378x and parameters in my oscam.conf file

in oscam.conf
[cache]

delay = 1
max_time = 14
max_hit_time = 3
csp_allow_request = 0
csp_allow_reforward = 0
wait_time = 350,0100:350,1702:1500,090F:900,0963:220 ,09CD:250,09C4:200,01:620,18:800,0B01:10 00,0D:1000,0500:600

i found somewhere wait_time left from capncook here in this forum

also reader and account in oscam.server and oscam.user for cacheex2 with cs378x

[reader]
label = label
protocol = cs378x
device = ip,port
user = xxx
password = xxx
group = # group here is unique
cacheex = 2
audisabled = 1
cacheex_maxhop = 10
cacheex_drop_csp = 1
cacheex_allow_request = 0

[account]
user = yyyyy
pwd = yyyyy
group = # all group here
cacheex = 2
cacheex_maxhop = 10
no_wait_time = 1
cacheex_drop_csp = 1
cacheex_allow_request = 0

in this acse in should make account with
user = xxx
pwd = xxx

and reader with your
device = ip,port
user = yyyyy
password = yyyyy

the cpu problem with 100% is solved

tested 2 days, till now i have doubled hits more from cache from before.

Pinky

[CyberCrime]

I will do some testing with your setting, but I get 90+% cache hit with my old settings and use about 7% cpu!
I guess the cpu problem as been solved, 2 weeks ago it was 80+% cpu

[CapNCooK]

Correct. HT solves CPU problem.

Anyways.. To get some more cache.. why not higher max_time..
It now does not influence CPU anymore, so put it to 60 or so.

[sat4ber]

@ CapNCook

Thanks for you're post. I've I understand you well Pinky needs to change this:

[cache]
delay = 1 > 0
max_time = 14 > 60
max_hit_time = 3 > 60
csp_allow_request = 0 > 1
csp_allow_reforward = 0 > 1wait_time = 350,0100:350,1702:1500,090F:900,0963:220,09CD:250,09C4:200,01:620,18:800,0B01:1000,0D:1000,0500:600 > OK?

[reader]
cacheex_drop_csp = 1 > 0

[account]
cacheex_maxhop = 10 >Maybe to much if you have a limit (upload,data or CPU) 
cacheex_drop_csp = 1 > 0

- Why not allow CSP? OScam can use it or not anymore?
'Also, in this HT, the whole 'group - loopback' discussion is also over. No more "one group", apply as you please ;)'

- So you must make for every cache share a unique group. Doesn't matter what for mode you are using? And what if you still using one group? Now you get so many groups.

- Do I not mis the CW Cycle Check?

[galiano]

..... In previous releases, group assignment mattered.. Not anymore.. Just use groups as you need them, it wont affect a thing.....

Which means that all cacheex peers by mpde-2, 3 can be in single group?

[CapNCooK]

// Reposted with more info //

@ CapNCook

In short..

- Proposed setting look fine to me
- In HT, push-IN does not affect cpu-load.., but push-OUT (still) does...
- Because of the above, limit all outgoing to cacheex_maxhop = 2 or max 3, and even better.. apply caid filters for your peers as required by them.
- For incoming cache, i would allow all.. cacheex_maxhop = 5 or so
- About CSP, sure it works, no reason to skip it..
- CWC is optional, but works.
- In previous releases, group assignment mattered.. Not anymore.. Just use groups as you need them, it wont affect a thing..

Few more hints:

About the csp_allow_* parameters.. these have only sense when you are connected to a CSP server obviously
If thats not the case.. leave them out.

About the waittimes set. Its not very known (also was not know by me..), but setting waittime in XX : XX : XX (CAID : STATIC_WAIT : DYN_WAIT) need to be set, especially if you use mode-1 cache! So your waittime settings, in optimal form without dynwait would be:

wait_time = 0:350:350,0100:350:350,1702:1500:1500,090F:900:900,0963:220:220,09CD:250:250,09C4:200:200,01:620:620,18:800:800,0B01:1000:1000,0D:1000:1000,0500:600:600

On a CacheEX-Only server, without any mode-1 users/readers.. wait_time can be set low. (since there are no hits on this server itself)
On a CacheEX-Only server, with mode-1 usage, just set wait_time to 60s.

[CapNCooK]

Which means that all cacheex peers by mpde-2, 3 can be in single group?

In pre-HT releases.. ONE single group had to be used.. not more.

Now, this requirement is completely gone..

[sat4ber]

Thanks!

Yes ,I understand the group assignment.

I put every cache user in one group, because I want to share everything with my users and then you keep your OScam file clean.

[galiano]

In pre-HT releases.. ONE single group had to be used.. not more.

Now, this requirement is completely gone..

Yes, but one single group for both modes is not wrong?

[CapNCooK]

Yes, but one single group for both modes is not wrong?

"Just use them as you need them".. => So no. it doesnt matter.

In HT, cache is now stored globally.. so, its impossible to have double-CW's in memory, regardless of any setting.