Max-pending exceeded

Hi,

I run CCcam since its first release but was forced to switch to oscam keeping my client connected with CCcam.
It works very stable with my local single cards.

My 3 identical 1702 cards on CCcam with loadbalancing enabled had perfect ECM timings of 600-1.500 ms.

Unfortunately Oscam doesn't have a good loadbalancer for identical cards to distribute the ECMs evenly to each reader knowing that each reader always responds in this constellation.

I tried out every loadbalancer mode, one card is always overloaded and then I have ECM timings rising > 5s causing timeouts.

The problem is that even with "lowest usage level" oscam sends multiple ECMs to one reader, when the reader does not respond within the retry limit the same ECM is sent to the second reader and sometimes also to the third and then the cards are overloaded and the system explodes.

CCcam does not do that, it sends each ECM only to one single card knowing it will respond sooner or later!

I tried to set up CSP only for my 3 identical 1702 cards with three newcamd-connectors to each oscam reader and one 1702 connection from oscam to CSP.
Now I have again perfect timings <600ms to each card and my 3 oscam readers have a 100% found-rate to the CSP-requests.

But some ECMs from my oscam where "not found" by the single 1702 newcamd-connection to the CSP.
I tried out every Filter and SID-assign on/of on both oscam an csp trying to figure out who filters or blocks the ECMs periodically,
cause it only happens periodically. My clients receive "not founds" on every 1702 channel and are not able to watch them without freezing.

My CSP filters about 20% of all 1702 ECM requests coming from oscam and does not even try to forward them to the 3 oscam-readers.
I tried every combination with "allow", "can decode" "default-max-queue" "max-queue" "betacrypttunnelplugin" and also betatunnel and ECM-Filters on/off on oscam but still have "not founds".

The CSP-log filters the ECMs with the reason "Max-pending exceeded":
[05:40:45.386] NewcamdSession[1:XXX_XX_beta] Filtered [Newcamd SEQ:291 SID:XX ECM (XXXX) 144 bytes] Reason: Max-pending exceeded

Maybe the estimated max-queue for each 1702 connector doesn't allow more forwards to the cards?

I cannot find any max-pending settings in the CSP-config.
I also tested max-cw-wait time of irrational 4.500 s instead of 4.500 ms in the 1702 profile giving an unlimited ECM-queue to each reader but the ECMs are still filtered!

I would be very happy if somebody can solve this problem, otherwise I have to run CCcam.x86_64 (2.1.3) only for my 1702 cards parallel to oscam.

Best regards

Chris

I have 3 identical sly de 1702 cards local on my Linux Server.

sync:
20% of the 1702 ECM request were not even forwarded by the csp to the oscam readers, the were filtered by the csp.
The forwarded ECMs were replied 100% by the three oscam readers, they have no SID-limitations activated.

As in my introduction mentioned I have a fanless Shuttle Slimline-PC with 3 Quadmouse readers:

XS35V3 Barebone
> Intel ATOM D2700 (2x2.13 GHz) Dual Core Prozessor
> 4 GB DDR3 SO-DIMM
> SATA III 60GB Intel 520 Serie SSD
> Slimline DVD Writer

I installed debian 6.0.6 amd64 with oscam, CCcam and CSP following the tutorials of this forum.

This is my oscam.server part for the 1702 setup:

[reader]
label = csp_1702
protocol = newcamd
device = 127.0.0.1,12200
key = 0102030405060708091011121314
user = osc1_1702
password = XXXXXXXX
disableserverfilter = 1
caid = 1702
ecmwhitelist = 93
ecmheaderwhitelist = 807090C7,817090C7
ident = 1702:000000
group = 1
audisabled = 1

[reader]
label = 1702-1
description = 1702-1
protocol = mouse
device = /dev/ttyUSB4
caid = 1702
boxkey = 1122334455667788
detect = cd
mhz = 600
cardmhz = 600
ident = 1702:000000
group = 11
emmcache = 1,3,2
dropbadcws = 1

[reader]
label = 1702-2
description = 1702-2
protocol = mouse
device = /dev/ttyUSB5
caid = 1702
boxkey = 1122334455667788
detect = cd
mhz = 600
cardmhz = 600
ident = 1702:000000
group = 12
emmcache = 1,3,2
dropbadcws = 1

[reader]
label = 1702-3
description = 1702-3
protocol = mouse
device = /dev/ttyUSB6
caid = 1702
boxkey = 1122334455667788
detect = cd
mhz = 600
cardmhz = 600
ident = 1702:000000
group = 13
emmcache = 1,3,2
dropbadcws = 1

musogeek: could you please move this thread to the csp support forum?

Sly de has up to 11 subchannels on Bundesliga (like your Premier League) and up to 13 subchannels on the Sports Channel.
The min. 1702 ECM time is 600 ms, the timeout is about 4.700 ms. I still don't understand when the time for viewing different channels is added, for example when users are viewing 7 different channels it can rise up to to 4.200 ms which is still ok, but also can stay between 600 - 1200 ms.

During normal usage two cards would be enough, but during scoccer when every user watches an other subchannel, 2 cards cannot always it.
I told you that my CCcam has ECM timings between 600 and 1.500 ms with three cards.

I read that when viewing for example 7 channels on different transponders the ECM time stays 600 ms as if they all were viewing one channel,
when viewing 7 channels on the same transponder all are added to a total of 4.200 ms whithout having timeouts.
So I distributed all 1702 channels for each 1702 transponder evenly in three oscam services lists for each card, the maximum was 3-4 channels/card on the biggest transponder.
Theoretically the ECM timings should be lower than 4*600ms = 2.400 ms, but they were not and oscam had only timeout with any loadbalancer mode.

The CCcam loadbalancer is unintelligent and has no learnig mode,
there is no retry after an unreplied ECM request, if the card does't answer, the client gets a NOK.
When I forgot to activate the 3D-Channel on one card, CCcam gave periodically NOK for the channel, when the request was forwarded to the card without 3D.
But for identical cards this stupid loadbalancing mode is perfect, because each card will always answer. No request is sent twice ore three times.

CSP distributes all requests evenly when having identical local cards just as CCCam does, when oscam requested for example 3.000 ECMs from the CSP (without the 20% filtered), CSP requested 1.000 from each oscam reader. Oscam without CSP connection does not manage that, it would have requested maybe 1.500 from one card and 1.500 from the two others together. I tried all modes, one card is always overloaded and there is unneccessary multiple identical ECM traffic to the cards.
You cannot predict if a ECM time is added when having queued ECMs to one card or not. So if Oscam sees one card has the lowest usage level it sends th next ECMs to that reader until the retry limit is reached but than it is maybe too late and this card timeouts, all ECMs are forwarded to the next cards which had also queued requests andalso timeouts and so on... like a chain reaction.

FYI: I have a subscription for a second card for only additional 10 EUR/month and use a friend's card as third to not ecounter problems when viewing soccer..

Sorry, I am new to this forum, this is my first thread after years and thought I would get professional support. I never needed help over the past 10 years for my cs and did everything by myself.
Why do you drop such hints? Why does anybody need a csp and you have a section here? I think we should not discuss policy matters but help each other. I thought my problem would be more interesting than just to explain "how to install x on y" or "why does emu x not read my card y?" and so on.

Does anybody really know about this csp log entry "Max-pending exceeded" when my oscam answers with "not found by csp_1702"? Should I post also my csp config?

Best regards

Chris

sync: CSP doesn't have the attribute "exclusive" for cannot-decode-services, only for can-decode where all unlisted SIDs are assumed not decodeable!
I started my testing only with a full 1702 can-decode exclusive list that should keep all services and then I also tried every combination of allow/block in the mapper section and can-decode /can-decode exclusive and cannot-decode in the cws-section.

It's not a mapper problem, otherwise CSP would return for example an empty reply with a NZE flag, that's what happened when I had all positive SID-lists and also the maper off!
Here's an extract of the log, when I tried it again this night:

[03:30:06.933] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:28 SID:204 ECM (0x80) 144 bytes]:
[03:30:07.096] NewcamdSession[96:***_de_beta] Sent [Newcamd SEQ:29 SID:0 DCW (0x80) 0 bytes]:
[03:30:07.097] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:29 SID:2b ECM (0x80) 144 bytes]:
[03:30:07.097] NewcamdSession[96:***_de_beta] Filtered [Newcamd SEQ:29 SID:2b ECM (0x80) 144 bytes] Reason: Max-pending exceeded
[03:30:07.199] NewcamdSession[96:***_de_beta] Sent [Newcamd SEQ:30 SID:0 DCW (0x81) 0 bytes]:
[03:30:07.199] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:30 SID:38 ECM (0x81) 144 bytes]:
[03:30:07.199] NewcamdSession[96:***_de_beta] Filtered [Newcamd SEQ:30 SID:38 ECM (0x81) 144 bytes] Reason: Max-pending exceeded
[03:30:07.493] NewcamdSession[96:***_de_beta] Sent [Newcamd SEQ:31 SID:0 DCW (0x81) 0 bytes]:
[03:30:07.493] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:31 SID:c ECM (0x81) 144 bytes]:
[03:30:07.493] NewcamdSession[96:***_de_beta] Filtered [Newcamd SEQ:31 SID:c ECM (0x81) 144 bytes] Reason: Max-pending exceeded
[03:30:07.494] NewcamdSession[96:***_de_beta] Sent [Newcamd SEQ:26 SID:8 DCW (0x81) 16 bytes]:
[03:30:07.495] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:32 SID:15 ECM (0x80) 144 bytes]:
[03:30:07.836] NewcamdSession[96:***_de_beta] Sent [Newcamd SEQ:33 SID:0 DCW (0x80) 0 bytes]:
[03:30:07.836] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:33 SID:2b ECM (0x80) 144 bytes]:
[03:30:07.836] NewcamdSession[96:***_de_beta] Filtered [Newcamd SEQ:33 SID:2b ECM (0x80) 144 bytes] Reason: Max-pending exceeded
[03:30:07.938] NewcamdSession[96:***_de_beta] Sent [Newcamd SEQ:34 SID:0 DCW (0x81) 0 bytes]:
[03:30:07.938] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:34 SID:38 ECM (0x81) 144 bytes]:
[03:30:07.939] NewcamdSession[96:***_de_beta] Filtered [Newcamd SEQ:34 SID:38 ECM (0x81) 144 bytes] Reason: Max-pending exceeded

In the status section of the CSP-WI "ECM filtered: " is about 20%.
I only left the logging plugin active and the ECMs were still filtered!
Which filters are active?

As you see in the log the filters occours after 18 seconds and 27 ECM requests on SEQ:29

SEQ:3 was the first ECM request from my oscam:
[03:29:49.640] NewcamdSession[96:***_de_beta] Recv [Newcamd SEQ:3 SID:81 ECM (0x81) 144 bytes]: