Opened 12 years ago

Closed 9 years ago

#2682 closed defect (expired)

Loadbalance (If a channel is not found it will not be cleared anymore)

Reported by: superg1972 Owned by: Micha_123
Priority: major Component: Loadbalancer
Severity: high Keywords:
Cc: Sensitive: no

Description

It looks like the loadbalance (lb_mode=1) doesn't consider the parameter to reopen the card readers because if you put a reopen fast or after time it doesn't change anything.It seems that with a timeout the card is completely forgotten for that channel. The reader with the timeout on a channel will not reopen anymore that channel...... you can wait days ( i tested it also) and the channel will be never opened.
As soon as you clear the timeout from the statistic via webif .... et voila' the channel is opened.

Here you can find an example of the issue:

If one channel of TVSat (183D, locaL CARD) is not found (and this is a strange issue that happens randomly, but few times)

183D:000000:0002:0000 Canale 5 008C found153 ms 123 ms6 2009.08.12 15:44:41
183D:000000:0CEA:0000 Rai Premium 008C found242 ms242ms 2 09.08.12 15:42:12
183D:000000:2140:0000 Rai 2 008C found 117ms122 ms 8 09.08.12 15:44:45
Not found
183D:000000:0004:0000 Iris 008C notfound 80100 ms 0 09.08.12 15:44:13

and if i go to that chanel on one client the image is not cleared because i get

2012/08/09 15:47:01 103742A0 c salotto (183D&000000/0000/0004/8C:FB38): rejected group (1 ms) - Iris (no matching reader)
2012/08/09 15:47:01 103742A0 c loadbalancer: NO MATCHING READER FOUND, reopen last valid:
2012/08/09 15:47:01 103742A0 c loadbalancer: reopened 0 readers
2012/08/09 15:47:01 103742A0 c loadbalancer: no best reader found, reopening other readers
2012/08/09 15:47:01 103742A0 c salotto (183D&000000/0000/0004/8C:FB38): rejected group (1 ms) - Iris (no matching reader)
2012/08/09 15:47:01 103742A0 c loadbalancer: NO MATCHING READER FOUND, reopen last valid:
2012/08/09 15:47:01 103742A0 c loadbalancer: reopened 0 readers
2012/08/09 15:47:01 103742A0 c loadbalancer: no best reader found, reopening other readers
2012/08/09 15:47:01 103742A0 c salotto (183D&000000/0000/0004/8C:FB38): rejected group (1 ms) - Iris (no matching reader)
2012/08/09 15:47:01 103742A0 c loadbalancer: NO MATCHING READER FOUND, reopen last valid:
2012/08/09 15:47:01 103742A0 c loadbalancer: reopened 0 readers
2012/08/09 15:47:01 103742A0 c loadbalancer: no best reader found, reopening other readers
2012/08/09 15:47:01 103742A0 c salotto (183D&000000/0000/0004/8C:FB38): rejected group (1 ms) - Iris (no matching reader)
2012/08/09 15:47:01 103742A0 c loadbalancer: NO MATCHING READER FOUND, reopen last valid:
2012/08/09 15:47:01 103742A0 c loadbalancer: reopened 0 readers
2012/08/09 15:47:01 103742A0 c loadbalancer: no best reader found, reopening other readers

As soon as i go in readers - Loadbalancer statistics - and i delete the line that is for the channel that is not found section (in this case Iris ) using the trash bin via webif, immediately the image is cleared and it works fine

2012/08/09 15:47:18 1024F7F0 h Reader tvsat stats 1 entry deleted by WebIF from XX.YY.XX.WW

2012/08/09 15:47:21 101A4CF8 r tvsat [nagra] Answer from cardreader:
2012/08/09 15:47:21 101A4CF8 53 14 01 00 00 00 8B C4 31 06 38 9E 4E 1E 62 B1
2012/08/09 15:47:21 101A4CF8 E3 01 09 53 02 A5 90 00
2012/08/09 15:47:21 101A4CF8 r tvsat [nagra] write to cardreader
2012/08/09 15:47:21 101A4CF8 A0 CA 00 00 8C D3 8A 00 00 00 00 00 79 DB EC 73
2012/08/09 15:47:21 101A4CF8 8E 64 07 F2 27 06 B5 47 FF D8 44 8B E7 08 7C 73
2012/08/09 15:47:21 101A4CF8 63 F4 2C 27 C9 13 31 D0 24 9D 65 1A ED 73 C7 59
2012/08/09 15:47:21 101A4CF8 1C 30 BE FF 55 35 69 3B 20 00 EF F0 5C 3A 3C 22
2012/08/09 15:47:21 101A4CF8 87 78 8C A9 7E 2B 88 08 3A 3C B3 9A F5 FE 92 F8
2012/08/09 15:47:21 101A4CF8 DA 14 64 10 13 86 B1 CF 9E F9 F6 3F BF 30 6E D2
2012/08/09 15:47:21 101A4CF8 36 E7 25 56 89 32 C7 4E 55 76 77 67 DF 47 1C 9D
2012/08/09 15:47:21 101A4CF8 51 9F 4B 85 68 59 07 B9 8F 30 91 A7 2C AD 34 F1
2012/08/09 15:47:21 101A4CF8 3A C0 96 AB 15 BA 22 4F 0C 78 1E 29 26 70 58 52
2012/08/09 15:47:21 101A4CF8 A5 16
2012/08/09 15:47:21 101A4CF8 r tvsat [nagra] Answer from cardreader:
2012/08/09 15:47:21 101A4CF8 53 14 01 00 00 00 FA D3 8C F4 4D 2C 64 EE 46 0D
2012/08/09 15:47:21 101A4CF8 5D C9 28 FC 3E F7 90 00
2012/08/09 15:47:21 103742A0 c loadbalancer: adding stat for reader tvsat: rc 0 183D&000000/0000/0004/8C:0875 time 181ms fail 0
2012/08/09 15:47:21 103742A0 c salotto (183D&000000/0000/0004/8C:0875): found (181 ms) by tvsat - Iris

I tried modifing a lot of settings without success.
The server is a PPC 7020. The version of Oscam is the last one (7245). If i remember well i started to have the issue from 70XX.

[global]
usrfile = /var/log/oscamuser.log
mailfile = /var/log/oscam.posta
logfile = /var/log/oscam.log
cwlogdir = /var/log/cw.log
emmlogdir = /var/log/emm.log
failbantime = 15
failbancount = 5
netprio = 1
nice = -20
maxlogsize = 1000
preferlocalcards = 1
readerrestartseconds = 7
block_same_ip = 0
block_same_name = 0
lb_mode = 1
lb_savepath = /var/log/lb.log
lb_reopen_mode = 1

[reader]
label = tvsat
protocol = internal
device = /dev/sci1
caid = 183D
boxid = XXXXXXXXX
rsakey = A9.....B5
ecmwhitelist = 8C
detect = cd
mhz = 500
cardmhz = 368
group = 1
lb_weight = 1000

[account]
user = salotto
pwd = XXXXXX
group = 1,2
au = tvsat
cccmaxhops = 2
cccreshare = 0
cccignorereshare = 1
keepalive = 1

Attachments (42)

foto.JPG (40.9 KB ) - added by superg1972 12 years ago.
Screenshot of the statistics in case of the issue
lb_tryfix1.patch (2.4 KB ) - added by theparasol 11 years ago.
LB tryfix1
lb_tryfix2.patch (6.4 KB ) - added by theparasol 11 years ago.
Attempt 2... more debug output and fixed failfactor
lbfix.patch (17.4 KB ) - added by blueven 11 years ago.
lbfix.patch
lbfix2.patch (24.0 KB ) - added by blueven 11 years ago.
lbfix3.patch (21.8 KB ) - added by blueven 11 years ago.
lbfix4.patch (29.7 KB ) - added by blueven 11 years ago.
lbfix4_2.patch (29.6 KB ) - added by blueven 11 years ago.
substitute for lbfix4.patch
lbfix5.patch (877 bytes ) - added by blueven 11 years ago.
Disable blocking readers with positive services defined. See ticket #3310, #3311
lbfix6.patch (827 bytes ) - added by blueven 11 years ago.
Fix lb_value=0 problem. It will be set to 1, so the reader will be not blocked.
lbfix7.patch (2.4 KB ) - added by blueven 11 years ago.
Fix asking too much time failed readers.
lbfix7_2.patch (9.2 KB ) - added by blueven 11 years ago.
replace for lbfix.patch
lbfix8.patch (5.9 KB ) - added by blueven 11 years ago.
lbfix9.patch (14.1 KB ) - added by blueven 11 years ago.
lbfix9rev8660.patch (10.8 KB ) - added by theparasol 11 years ago.
LB9fix diff patch for 8660 (orginal lb9fix of blueven is against rev 8658)
ecm_lb9fix.patch (41.5 KB ) - added by blueven 11 years ago.
New patch with a lots of changes, including lbfix9
ecm_handler_2.patch (32.0 KB ) - added by blueven 11 years ago.
tryfix 2 for ecm handler. Tested on #8664
ecm_handler_4.patch (45.3 KB ) - added by blueven 11 years ago.
ecm handler 4 (patch)
ecm_handler_5.patch (45.4 KB ) - added by blueven 11 years ago.
ecm_handler_5-2.patch (46.8 KB ) - added by blueven 11 years ago.
tryfix writelock/readlock problem.
ecm_handler_5-3.patch (48.8 KB ) - added by blueven 11 years ago.
fix cache stats
ecm_handler_6.patch (59.2 KB ) - added by blueven 11 years ago.
Make accessing ecm request by cacheex code "thread safe".
ecm_handler_7.patch (65.9 KB ) - added by blueven 11 years ago.
ecm_handler_8.patch (60.2 KB ) - added by blueven 11 years ago.
ecm_handler_9.patch (64.2 KB ) - added by blueven 11 years ago.
ecm_handler_9-1.patch (65.3 KB ) - added by blueven 11 years ago.
ecm_handler_9-2.patch (65.9 KB ) - added by blueven 11 years ago.
ecm_handler_10.patch (86.7 KB ) - added by blueven 11 years ago.
Deatils on forum
ecm_handler_10-1.patch (86.9 KB ) - added by blueven 11 years ago.
fix bug from v10
ecm_hanlder_11.patch (89.1 KB ) - added by blueven 11 years ago.
Details on forum
ecm_handler_11-1.patch (96.1 KB ) - added by blueven 11 years ago.
ecm_handler_11-2.patch (97.3 KB ) - added by blueven 11 years ago.
ecm_handler_11-3.2.patch (103.0 KB ) - added by blueven 11 years ago.
ecm_handler_11-3.patch (103.0 KB ) - added by blueven 11 years ago.
ecm_handler_11-4.patch (106.0 KB ) - added by blueven 11 years ago.
ecm_handler_11-5.patch (112.0 KB ) - added by blueven 11 years ago.
ecm_handler_11-6.patch (113.7 KB ) - added by blueven 11 years ago.
ecm_handler_pacth_11-6.1.patch (114.1 KB ) - added by blueven 11 years ago.
fix for 11-6 patch
ecm_handler_11-7.patch (114.8 KB ) - added by blueven 11 years ago.
ecm_handler_11-8.patch (117.0 KB ) - added by blueven 11 years ago.
ecm_handler_11-9.patch (119.6 KB ) - added by blueven 11 years ago.
ecm_handler_12.patch (8.8 KB ) - added by blueven 11 years ago.

Change History (118)

comment:1 by superg1972, 12 years ago

I forgot to report that with version 7250 the issue persists.

comment:2 by superg1972, 12 years ago

I tested with the last versions and the issue persists. If a channel has gone in the not found list (with a standard ms ECM also)and so i get the no matching reader (channel not cleared out) and i put lb_mode=10 or i delete the statistics for the loadbalancer or from the readers the channel is immediately cleared out.

comment:3 by testdrive, 12 years ago

I have the same problem, using x86, started to happen after version 70XX. It's usually happening within 24 hours after server start.
Version 7266, still same issue.

comment:4 by corsair, 12 years ago

Resolution: fixed
Status: newclosed

Yes, I think this happens if you have no services on your local cards defined.
I found a bug and fixed it with r7277. Please test and report.

comment:5 by superg1972, 12 years ago

I'll test it, but this happens with and without services on the local reader

Last edited 12 years ago by superg1972 (previous) (diff)

comment:6 by superg1972, 12 years ago

Resolution: fixed
Status: closedreopened

Unfortunately the issue has not been fixed. I waited to answer to check better the situation, but i got the same issue. It happens on readers with and without services and
with lb_reopen_mode = 1 and lb_reopen_mode = 0.
A workaround that i have to apply is to delete the statistics every hour, but unfortunately some times if the timeout is renewed it is not deleted and 1 hour is a very long time to wait.
I think that if the card is local the timeouts should not be evere taken in consideration for the loadbalance.

comment:7 by corsair, 12 years ago

You can avoid blocking by setting services on your local cards.
Also with lb_reopen_seconds you can modify the reopen time after a blocking.

lb_reopen_seconds is the key to reopen a failing channel. But if ist failing too often, a fail-factor is increased. And this fail-factor is a multiplicator for lb_reopen_seconds.

comment:8 by corsair, 12 years ago

Also, lb_reopen_mode=1 should ALWAYS reopen a channel, if no reader ist found

comment:9 by superg1972, 12 years ago

Unfortunately this is not the case

comment:10 by ni hao, 12 years ago

tested here, no problem with LB reopen mode 1. Although I do not think it makes sense for you, but try to use valid lb_weight values and not invalid. 1000 is invalid according to wiki.

Version 3, edited 12 years ago by ni hao (previous) (next) (diff)

comment:11 by superg1972, 12 years ago

The issue is still present in the last versions.

I spoke with other people and they were reporting me the fact that they were loosing the card for some channel. I checked with them and they had the channel as Not found in the statistics. As soon as they deleted the channel from the not founds, the channel reappeared.

comment:12 by badjoker, 12 years ago

Priority: minormajor
Severity: mediumhigh

I can confirm the same and it is really annoying. I just try to find the changeset, which could be cause this bug, maybe it is this one?

http://www.streamboard.tv/oscam/changeset/7040

I'll try right now and give feedback.

It will block all cards after some time of running, this way no loadbalancing is possible at all!

The weird thing about it is, the cards that are unique are not affected at all in the same server. So I run 3 cards of one provider that are identical, they show this problem and in the same server there are 6 different unique cards that one's are not affected at all.

Feedback: OK, 1.20 rev 7039 does have this bug too. I will keep on testing older versions!

Last edited 12 years ago by badjoker (previous) (diff)

comment:13 by pcz, 12 years ago

I'm faced with this bug too. Unfortunately the issue has not been fixed in 7427 as well :(. I'm very interesting, which is the last one, which we can use without this bug. And as well, this isn't a new bug, why isn't solved it by team?

by superg1972, 12 years ago

Attachment: foto.JPG added

Screenshot of the statistics in case of the issue

comment:14 by nocion, 11 years ago

nothing yet??, on 7775 bug still persists

Last edited 11 years ago by nocion (previous) (diff)

comment:15 by TNe-TWo, 11 years ago

@superg1972
it is possible to get a log from oscam build with debug on

we need more info like
2012/11/02 07:18:11 195A430 h loadbalancer: client ecmclient for 1834&000000/0000/0015/89:0000: n=1 valid readers: ttyusb0L
2012/11/02 07:18:11 195A430 h loadbalancer: NO MATCHING READER FOUND, reopen last valid:
2012/11/02 07:18:11 195A430 h loadbalancer: reopened 0 readers
2012/11/02 07:18:11 195A430 h loadbalancer: no best reader found, reopening other readers
2012/11/02 07:18:11 195A430 h loadbalancer: reader ttyusb0 need 155 seconds to reopen
2012/11/02 07:18:11 195A430 h loadbalancer: client ecmclient for 1834&000000/0000/0015/89:0000: n=0 selected readers:

comment:16 by ciberoscam, 11 years ago

I subscrive this post too, its a problem that i myself face at much time ago, so i gived up loadbalancer, with good configurations you dont need it and you can even improve performance, athought i get some configuration working +/- fine for loadbalancer:

lb_mode = 3
lb_max_ecmcount = 150
lb_stat_cleanup = 12

Each 150 ECMs for that reader it uses again the card for timeout chanels, each 12 hour he cleans statistics, try it and change it how do you like.

comment:17 by Micha_123, 11 years ago

Same here, im using the old 5766 stable osacam on my E3000 with DDwrt. in that version is it nice working but i'm tested "OSCAM 1.20-unstable_svn build #6789 " but it dosn"t work.

i dont have services defined and it is not the solution. if a channel is Timed out so it must be reopend with and without defined services after reopen time or not ?

comment:18 by Micha_123, 11 years ago

Component: GeneralLoadbalancer
Owner: set to Micha_123
Status: reopenednew

comment:19 by blueven, 11 years ago

Same problem here. If a reader is block for timeout or not found, it will be NO cleared if no reader is found for the same channel. So not found is given to clients request that channell.

Why no one want resolve this very old and annoying bug?

Last edited 11 years ago by blueven (previous) (diff)

by theparasol, 11 years ago

Attachment: lb_tryfix1.patch added

LB tryfix1

comment:20 by theparasol, 11 years ago

Does this patch change something for the better?

Hourly houskeeping -> never worked due to initial lasthouskeeping = 0 -> fixed!
Failfactor could run sky high in no time 2*2*2... now its 1,2,3,4,5 * reopenseconds
Re-added reopen reader in case not found (disabled since rev 6380)

comment:21 by blueven, 11 years ago

At line 1342 we should also check cfg.lb_reopen_mode=1? When "NO MATCHING READER FOUND" we should unblock reader in "reopen fast", if set, imho. Do you agree?

if (s && s->rc != E_FOUND && s->last_received+get_reopen_seconds(s) < current_time) {

change with:

if (s && s->rc != E_FOUND && (s->last_received+get_reopen_seconds(s) < current_time || cfg.lb_reopen_mode)) {

p.s. I think the problem is from line 1335. In this section of code, it should reopen readers when no best reader is found, but something does not work. Your uncomment section should reopen readers even when best reader is found. Both fix are usefull imho.

I'll try it tommorrow.

Thanks

Last edited 11 years ago by blueven (previous) (diff)

comment:22 by theparasol, 11 years ago

Dont think so, its already covered by 1358.

Updated the fix again, failfactor didnt work and stayed always on 1 fail time.

by theparasol, 11 years ago

Attachment: lb_tryfix2.patch added

Attempt 2... more debug output and fixed failfactor

comment:23 by blueven, 11 years ago

Why we should reopen readers only when no best reader found?
I think your first patch is better, when reader reaches reopen seconds, we must reopen it for try again, not important best reader is found or not.
However i'm also working on the code, i will give you my patch so you can observe if my changes are applicable also in your patch.

comment:24 by theparasol, 11 years ago

Why open a reader that is notorious for failing while there is a up and running steady reader makes no sense. What could happen is that the failing reader gets reopened again and is going down again right away. This is something you dont want.
While now the inactive readers will get reopened if there is really a reason to do so.

If I reenable the remarked code in tryfix2 it will make the code from 1358 obsolete, so dont doing that without a good reason...

comment:25 by blueven, 11 years ago

I understand what you are speaking about... but if a reader give me for some period of time "not fund" due anticascading penalty, or if a reader is in timeout due network or dns problem... why cannot retry this reader after some time? It could be my best reader, but it is down at some moment for any reason. In this way, i should wait all my readers go down or i reset statistics for reopen it! This is something i dont want.
Furthermore, fail_factor will prevent this reader is reopened too often. If it fails consecutively, the time for reopening will increase, so the reopening will not be invasive.
While, if no "NO MATCHING READER FOUND", it is important that readers are reopened immediately or after reopen_seconds, depending how we set lb_reopen_mode.

I'm cleaning and fixing "stat_get_best_reader" function, according to what is right for me and based on your first patch. I'm testing it. If it is good, tomorrow i will send the patch so you can tell me if my changes are valid or not.

Last edited 11 years ago by blueven (previous) (diff)

by blueven, 11 years ago

Attachment: lbfix.patch added

lbfix.patch

comment:26 by blueven, 11 years ago

Hi, here my patch.
I test it some time, and it works very well. I'm testing more.

These are changes:

  1. Not used anymore lb_reopen_mode (it will run always in 0 mode--> no need for reopen_fast!)
  2. If a reader reaches reopen_seconds*fail_fatcor, cleaning stats (only) for it and reopening (only) it even if we have valids readers. If it fails consecutively, it will block more time due fail_factor.
  3. If avg_time > retrylimit, cleaning stats and reopening all matching readers.
  4. If NO MATCHING READER FOUND, cleaning stats and reopening all matching readers.
  5. if lb_max_readers is not 0, and lb_max_readers<=(nbest_readers+nfb_readers), it will be set at (nbest_readers+nfb_readers)+1.
  6. If ecm_count > lb_max_ecmcount, cleaning stats for the reader (only it), without check retrylimit is over or not!
  7. Add fix for fail_factor (1,2,3,4,5..*reopen_seconds) and housekeeping_stat by theparasol.
  8. more debug output

comment:27 by blueven, 11 years ago

I noted there is a problem with lb_max_readers>0. No problem with lb_max_readers=0.
New patch is coming, i'm fixing it.

comment:28 by Mo0211, 11 years ago

Hi blueven,

your patch works very well!
thank you and parasol for contributing this :)

I'm waiting for the new version...

comment:29 by littlejoe, 11 years ago

Hi,

Any chance that this patch will be implemented soon?

Thanks!

comment:30 by Mo0211, 11 years ago

with the newest-Version (lbfix.patch) and the actual rev 8610 i get this error-message:

patch -p0 <lbfix.patch
(Stripping trailing CRs from patch.)
patching file module-stat.c
patch unexpectedly ends in middle of line
Hunk #18 succeeded at 1397 with fuzz 1.

Is something wrong with the patch?

comment:31 by today, 11 years ago

You must rename:

+++ module-stat_patched.c
to
+++ module-stat.c

comment:32 by blueven, 11 years ago

Hi, thanks for testing!
In this weekend i hope release the new patch. I'm still working on it.
After, i would introduce the possibility to set nbest_reader, nbest_reader_percaid and max_readers for user, so we can handle lb differently for any user.

by blueven, 11 years ago

Attachment: lbfix2.patch added

comment:33 by blueven, 11 years ago

Hi, here the new patch.
These are overall changes (included first patch):

  • Fix preferlocalcards. If preferlocalcards=1, we try to use always local reader first, even when we try to reopen readers.
  • Fix for lb_max_readers!=0. Now, before try to reopen readers, we are sure first select readers with last ecm "FOUND"!
  • If best avg_time > retrylimit, cleaning stats and force reopening all matching readers.
  • If NO MATCHING READER FOUND, cleaning stats and force reopening all matching readers.
  • Not used anymore lb_reopen_mode (it will run always in 0 mode--> no need for reopen_fast!), using reopen_seconds.
  • If a reader reaches reopen_seconds, cleaning stats for it and reopening it. If it fails consecutively, it will block more time due to fail_factor.
  • Fix considering lb_weight in lb_mode=2
  • (Re-introduced) Reset stats when ecmcount>lb_max_ecmcount only when its time_avg > retrylimit.
  • Blocked readers are readers with last ecm not "found" (blocked for reopen_secods*fail_factor), and readers with ecm_count>lb_max_ecmcount and avg-time>retrylimit.
  • If lb_max_readers is not 0, and lb_max_readers<=nbest_readers, it will be set at nbest_readers+1.
  • lb_max_readers only affects nbest_readers and blocked readers for reopening. Example: we set nbest_readers=2 and lb_max_readers=5, we try to select 2 nbest_readers and try to open 3 blocked readers.

No limit in fallback readers number.
No limit readers need starting stats. (they will be always activated)
No limit readers not reach lb_min_ecmcount. (they will be always activated)

  • Add fix for fail_factor (1,2,3,4,5..*reopen_seconds) and housekeeping_stat by theparasol.
  • more debug output

comment:34 by blueven, 11 years ago

Please, try patch and report any problem.
Thanks

comment:35 by Mo0211, 11 years ago

hi, again this error message:

patch -p0 < ../lbfix2.patch
(Stripping trailing CRs from patch.)
patching file module-stat.c
patch unexpectedly ends in middle of line
Hunk #24 succeeded at 1459 with fuzz 1.

comment:36 by mumbua, 11 years ago

Patch(e) anwenden lbfix2.patch â

â (Stripping trailing CRs from patch.) â
â patching file module-stat.c â
â patch unexpectedly ends in middle of line â
â Hunk #24 succeeded at 1459 with fuzz 1.

I try R8618,8617,8616,8615
Its work with Revision ??

comment:37 by amarian12, 11 years ago

just edit the patch file and add extra empty line at end of the files will work perfect with latest version of oscam.

patch -p0 < ../lbfix2.patch --verbose
Hmm... Looks like a unified diff to me...
The text leading up to this was:


module-stat.c 2013-04-15 02:03:58.841768745 +0200

|+++ module-stat.c 2013-04-15 02:03:19.422264314 +0200


Patching file module-stat.c using Plan A...
Hunk #1 succeeded at 12.
Hunk #2 succeeded at 383.
Hunk #3 succeeded at 432.
Hunk #4 succeeded at 463.
Hunk #5 succeeded at 505.
Hunk #6 succeeded at 540.
Hunk #7 succeeded at 608.
Hunk #8 succeeded at 639.
Hunk #9 succeeded at 659.
Hunk #10 succeeded at 667.
Hunk #11 succeeded at 737.
Hunk #12 succeeded at 837.
Hunk #13 succeeded at 853.
Hunk #14 succeeded at 1114.
Hunk #15 succeeded at 1166.
Hunk #16 succeeded at 1212.
Hunk #17 succeeded at 1225.
Hunk #18 succeeded at 1234.
Hunk #19 succeeded at 1260.
Hunk #20 succeeded at 1275.
Hunk #21 succeeded at 1284.
Hunk #22 succeeded at 1376.
Hunk #23 succeeded at 1407.
Hunk #24 succeeded at 1459.
done

comment:38 by blueven, 11 years ago

Hi, i have no problem using my patch with latest revision.
I only do: patch -p0 < lbfix2.patch
patch file and module-stat.c are in the same dir.

Last edited 11 years ago by blueven (previous) (diff)

comment:39 by mumbua, 11 years ago

just edit the patch file and add extra empty line at end of the files will work perfect with latest version of oscam.

Yes is working

comment:40 by duckman, 11 years ago

Please add it to the trunk!

by blueven, 11 years ago

Attachment: lbfix3.patch added

comment:41 by blueven, 11 years ago

Hi, here my new patch, hope latest if no bug exist.

These are overall changes (included first patch):

  • Fix preferlocalcards. If preferlocalcards=1, we try to use always local reader first.
  • Fix for lb_max_readers!=0. Now, before try to reopen readers, we are sure first select readers with last ecm "FOUND"!
  • If best avg_time > retrylimit, cleaning avg-time and force reopening all matching readers.
  • If NO MATCHING READER FOUND, force reopening all matching readers.
  • Not used anymore lb_reopen_mode (it will run always in 0 mode--> no need for reopen_fast!), using reopen_seconds.
  • If a reader reaches reopen_seconds, reopening it. If it fails consecutively, it will block more time due to fail_factor. In detail, if at reopen it does not get lb_min_ecmcount "found" consecutively, fail_factor will be increase. Otherwise, fail_factor will be reset.
  • Fix considering lb_weight in lb_mode=2
  • (Re-introduced) Reset stats only when ecmcount>lb_max_ecmcount and time_avg > retrylimit.
  • If lb_max_readers is not 0, and lb_max_readers<=nbest_readers, it will be set at nbest_readers+1.
  • lb_max_readers only affects nbest_readers and blocked readers for reopening. Example: we set nbest_readers=2 and lb_max_readers=5, we try to select 2 nbest_readers and try to open 3 blocked readers.

No limit in fallback readers number.
No limit readers need starting stats. (they will be always activated)
No limit readers not reach lb_min_ecmcount. (they will be always activated)
No limit readers reach lb_max_ecmcount and time_avg > retrylimit. (they will be always activated)

  • Add fix for fail_factor (1,2,3,4,5..*reopen_seconds) and housekeeping_stat by theparasol.
  • more debug output
  • Add two more check:
    1. A reader is considered VALID for lbvalue calculation , if it reaches lb_min_ecm_count. But now, this check is made not only at starting stats, but also when a reader is reopened after block and when oscam restart! It is usefull for re-evaluating avg time and for being sure reader is now valid after a block!
    2. If a reader fails consecutively, fail_factor is incresed. But this should not happen when we force to reopen the reader when no matching reader found! So, fail factor will be increase each reopen_seconds*fail_factor.
  • Add lb_nbest_readers and lb_nbest_percaid parameters recognition in oscam.user.

You should delete your stats before starting oscam, because the format is changed, i added an other value in stat file.

For apply the patch, put the file in oscam root dir and run:

patch -p0 -i lbfix3.patch

Tested with last revision #8634.

comment:42 by FilipeAmadeuO, 11 years ago

@blueven, many thanks for the patch.
Can you remove not used anymore parameters ? Reopen mode and Time to reopen (i think both are not used anymore)
Thanks

by blueven, 11 years ago

Attachment: lbfix4.patch added

comment:43 by blueven, 11 years ago

Changes for lbfix4.patch:

  • VERY IMPORTANT FIX: detect cw=0 as not_found in oscam check, and LB treats it as not found (in log stat it is stamped as rc 7, fake rc!)
  • Set "dummy" avg time to 0ms, no more 80000ms. Cosmetic issue!
  • Remove reopen_mode from code and webif.
  • No more "Log statistics only" mode.
  • No more reset avg-time when a reader is reopened. No need to do it for better tuning. Reset avg-time only when retrylimit reached.

by blueven, 11 years ago

Attachment: lbfix4_2.patch added

substitute for lbfix4.patch

comment:44 by blueven, 11 years ago

Does not consider anymore lbfix4.patch.
Last valid patch is lbfix4_2.patch​.

Fix fake cw=0. Now they are saving as not_found in lb stats. No broke anymore lb stats for "real" not_found.

comment:45 by theparasol, 11 years ago

Since this latest patch works better than previous ones I committed it to the trunk

comment:46 by malakudi, 11 years ago

What is the purpose of lb_nbest_* in oscam.user? Configuring load balancer differently per user? For what benefit?

For best performance of the load balancer, I think we also need clienttimeout and fallbacktimeout values defined per caid. For NDS, we need fallbacktimeout lower than 450 ms, for other CAIDs we might need fallbacktimeout to 2000ms or higher. What do you think?

comment:47 by blueven, 11 years ago

Yes, you read my thoughts!!! :)
I'm valuating about fallbacktimeout_percaid.

Nonetheless, lb_nbest_* in oscam.user could be usefull for differentiating LB for user.

comment:48 by 123123123, 11 years ago

Is it possible, that lb_nfb_readers isn't considered anymore on fallback?
Because i see timeouts from fallback in log with requests to up to lb_max_readers and not just to lb_nfb_readers.

comment:49 by blueven, 11 years ago

Hi, i cannot answer you without seeing log -d256. If you notice something wrong, attach the log.

by blueven, 11 years ago

Attachment: lbfix5.patch added

Disable blocking readers with positive services defined. See ticket #3310, #3311

comment:50 by 123123123, 11 years ago

I have to correct my report - on fallback all readers are opened. Not only lb_nfb or lb_max_readers.
I have set lb_nfb_readers = 2 and also lb_max_readers = 2

But on fallback:

2013/05/13 22:46:23 77511030 c loadbalancer: client api for 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD: n=6 valid readers: serverA serverB serverC serverD serverE ...(1 more)
2013/05/13 22:46:23 77511030 c loadbalancer: --------------------------------------------
2013/05/13 22:46:23 77511030 c loadbalancer: nbest 1, nfb 2, max_reopen 1
2013/05/13 22:46:23 77511030 c loadbalancer: reader serverA need starting statistics --> ACTIVE
2013/05/13 22:46:23 77511030 c loadbalancer: reader serverB need starting statistics --> ACTIVE
2013/05/13 22:46:23 77511030 c loadbalancer: reader serverC need starting statistics --> ACTIVE
2013/05/13 22:46:23 77511030 c loadbalancer: reader serverD need starting statistics --> ACTIVE
2013/05/13 22:46:23 77511030 c loadbalancer: reader serverE need starting statistics --> ACTIVE
2013/05/13 22:46:23 77511030 c loadbalancer: reader serverF need starting statistics --> ACTIVE
2013/05/13 22:46:23 77511030 c loadbalancer: --------------------------------------------
2013/05/13 22:46:23 77511030 c loadbalancer: client api for 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD: n=6 selected readers: serverA serverB serverC serverD serverE ...(1 more)
2013/05/13 22:46:26        0   loadbalancer: adding stat for reader serverA: rc 5 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD time 3850ms fail 1
2013/05/13 22:46:26        0   loadbalancer: adding stat for reader serverB: rc 5 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD time 3851ms fail 1
2013/05/13 22:46:26        0   loadbalancer: adding stat for reader serverC: rc 5 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD time 3851ms fail 1
2013/05/13 22:46:26        0   loadbalancer: adding stat for reader serverD: rc 5 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD time 3852ms fail 1
2013/05/13 22:46:26        0   loadbalancer: adding stat for reader serverE: rc 5 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD time 3852ms fail 1
2013/05/13 22:46:26 77511030 c api (09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD): timeout (3851 ms) by serverA,serverB,serverC,serverD,serverE,serverF (F/6/6/6)
2013/05/13 22:46:26        0   loadbalancer: adding stat for reader serverF: rc 5 09CD&000000/0000/38BE/B7:9856166D3C6281CCD361F9FFEDC3C6FD time 3853ms fail 1

You see, all available readers are asked on timeout. On Fallback stage..

On my Server i see all the time things like:
(max set to 8 and nfb to 1 - log filtered for a SID, where i get many requests which lead to a timeout: "0262")

Server(filtered): [LOG690]2013/05/13 23:08:47 A001A710 c UserA (0B00&000000/0000/0262/67:D049267EA49394489A695AF4A003A327): timeout (4635 ms) by ServerA,ServerB,ServerC,ServerD,ServerE,ServerF,ServerJ,ServerG,ServerH (F/31/31/31) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262
Server(filtered): [LOG691]2013/05/13 23:08:47 A001A710 c UserA (0B00&000000/0000/0262/67:D049267EA49394489A695AF4A003A327): timeout (3801 ms) (F/0/31/31) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262
Server(filtered): [LOG692]2013/05/13 23:08:47 A41FAAB0 c UserB (0B00&000000/0000/0262/67:D049267EA49394489A695AF4A003A327): timeout (4447 ms) (F/0/30/30) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262
Server(filtered): [LOG589]2013/05/13 23:09:21 A41FAAB0 c UserB (0B00&000000/0000/0262/67:519A4BE0C8BA9888D5B8CBE79E00A22F): timeout (4549 ms) by ServerA,ServerB,ServerC,ServerD,ServerE,ServerF,ServerJ,ServerG,ServerH (F/30/30/30) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262
Server(filtered): [LOG854]2013/05/13 23:09:32 A001A710 c UserA (0B00&000000/0000/0262/67:09A37FFDA7EE795CED45879CFCE98E5A): timeout (4504 ms) (F/0/31/31) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262
Server(filtered): [LOG856]2013/05/13 23:09:32 A41FAAB0 c UserB (0B00&000000/0000/0262/67:09A37FFDA7EE795CED45879CFCE98E5A): timeout (4777 ms) by ServerA,ServerC,ServerD,ServerE,ServerF,ServerJ,ServerG,ServerH,ServerI (F/30/30/30) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262
Server(filtered): [LOG251]2013/05/13 23:09:47 A001A710 c UserA (0B00&000000/0000/0262/67:51837F21216CAC28791F3E975EE233CF): timeout (4501 ms) by ServerB,ServerC,ServerD,ServerE,ServerF,ServerJ,ServerG,ServerH, (F/31/31/31) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262
Server(filtered): [LOG252]2013/05/13 23:09:47 A41FAAB0 c UserB (0B00&000000/0000/0262/67:51837F21216CAC28791F3E975EE233CF): timeout (3996 ms) (F/0/30/30) - caid: 0B00 ident: 000000 pid: 0000 sid: 0262

I'd expect that it is max. lb_max_readers or even lb_nfb_readers (if there are some in stats) - but not all available readers..

Or am i thinking totally wrong?

Last edited 11 years ago by 123123123 (previous) (diff)

comment:51 by blueven, 11 years ago

When "lb_retrylimit reached" or "no valid matching readers" all available readers are used, like oscam without loadbalance.

You see all readers are used because no valid matching reader is found. However, for these questions, we should use the forum.

by blueven, 11 years ago

Attachment: lbfix6.patch added

Fix lb_value=0 problem. It will be set to 1, so the reader will be not blocked.

by blueven, 11 years ago

Attachment: lbfix7.patch added

Fix asking too much time failed readers.

comment:52 by blueven, 11 years ago

lbfix7 try to fix asking a lot of times failed readers.
Now, when no matching reader found and when need to reset avg time for lb_retrylimit, we force to reopen only ONE time readers with min_ecm_count reached, setting fail_factor to 0.
If it fails again, we increase fail_factor (0->1) and we retry to reopen it only after reopen_seconds*fail_factor. And so on.
Knock failed reader only one time.

Default reopen_seconds to 30 seconds.

For readers with ecm_count<min_ecm_count, we don't knock it.

Please try and report in the forum.

Last edited 11 years ago by blueven (previous) (diff)

by blueven, 11 years ago

Attachment: lbfix7_2.patch added

replace for lbfix.patch

comment:53 by blueven, 11 years ago

lbfix7_2.patch

  • Replace for lbfix.patch (fix fail_factor reset on knocking)
  • add new parameter "lb_whitelist_services" in reader configuration. Now, positive services ("services" parameter) are handled normally. If someone want services that are whitelist for lb, could set positive (only positive, negative does not have effect) services in new parameter. This parameter is what users in ticket #3310,#3311 would use.


E.g.
lb_whitelist_services=<services>[,<services>]....

@theparasol: revert latest patch (lbfix7) and apply only this. Thanks

Last edited 11 years ago by blueven (previous) (diff)

comment:54 by Mo0211, 11 years ago

do i need to specify the normal services or the caid in addition?

Because if i configure a camd35 reader with whitelisting 1702-services every possible caid will be asked.

Is this the normal behavior?

comment:55 by blueven, 11 years ago

Yes, services in "lb_whitelist_services" parameter caused reader will never blocked by loadbalancer even if it fail.
Services in "services" parameter are handled normally, no special handling.

by blueven, 11 years ago

Attachment: lbfix8.patch added

comment:56 by blueven, 11 years ago

lbfix8 changes:

  • Some code cleaning.
  • Remove check on min ecm count consecutively when oscam re-start and after retrylimit reached. His benefit is not so good compared more trafiic generated.
  • Improve handle retrylimit in mode 2 and mode 3. Retrylimit check in mode 2/3 not has many sense, because we select readers not using avg time. Currently, retrylimit check could generate imprevedible selection. So, i improved retrylimit check in mode 2/3. Now, it simply checks avg-time during reader selection, and if reader has avg-time>retrylimit, lb_value will be set to the smallest value. In this way reader selection will never use "over-limit" readers until we have "under-limit" readers. Only when all valid readers reach retrylimit, the "standard" retrylimit check will be called: reset all avg time, activate all readers and "knocking" blocked readers (not knocked yet).
Last edited 11 years ago by blueven (previous) (diff)

comment:57 by Winz, 11 years ago

lbfix8 works very good on my lb_mode2 plz add this patch in the trunk
good job blueven

by blueven, 11 years ago

Attachment: lbfix9.patch added

comment:58 by blueven, 11 years ago

lbfix9.patch changes:

  • Add check for better tuning on retrylimit: if a reader has time-avg>retrylimit and last received ecm older than 5 minutes, it will be forced to "active" and avg-time resets. So, it will be re-evaluated and no waiting anymore when (and if) all readers reaches retrylimit for re-evaluated it.
  • lb_retrylimit = 0 as default. When zero, no checks on retrylimit will be done. This is important because previous 800ms is no a good deafult value for all. So people needed it, will personalize the value.
  • Remove last_inc_failfactor: no more needed
  • Add lb_nfb_readers in user config
  • Dummy avg time is now 99999. It is usefull to prevent the reader get selected while "knocking" it.

IMPORTANT: delete stat file because not compatible with previous.

by theparasol, 11 years ago

Attachment: lbfix9rev8660.patch added

LB9fix diff patch for 8660 (orginal lb9fix of blueven is against rev 8658)

comment:59 by mumbua, 11 years ago

I patch Revision 8558 and now 8660. Hammering is in both Revision over.
Courious is i dont hammer all my Peers. But one Peer i hammer extremly. Has goen back to normal now.From my Side is ready for Trunk.

comment:60 by blueven, 11 years ago

If one Peer is hammering extremly, it depends on your lb settings and on ecm-time of yours readers if you use retrylimit.

Last edited 11 years ago by blueven (previous) (diff)

by blueven, 11 years ago

Attachment: ecm_lb9fix.patch added

New patch with a lots of changes, including lbfix9

comment:61 by blueven, 11 years ago

ecm_lb9fix.patch changes:

  • Immediately call to fallback readers when all "not fallback" readers already answered. So, we don't have to wait fallbacktimeout if we there aren't others answers to wait!
  • Add fallbacktimeout_percaid parameter.
  • Preciser ecm-time computation for LB stats!
  • Not call LB handler for ecm in cache. Fix it.
  • move check about cw=0 in write_ecm_answer function, instead of chk_dcw function.
  • small fixs and code optimization/cleaning in write_ecm_answer.
  • small fixs and code optimization/cleaning in chk_dcw.
  • send stats only in chk_dcw that is thread safe!
  • callback for wait_time_cacheex, fallbacktimeout and clienttimeout are made now in thread-safe manner!!!
  • A reader got timeout imemdiately after clienttimeout if not answer, even if cw for ecm is "found". Not wait other ecm to add it to stats. Fix it.
  • In distribute_ecm for cache anwsers, we have to not call again wrtire_ecm_answer, we have to send cw/answer to client!

The patch include latest lbfix9 patch. I generated it on #8662 revision.

Please test and report.

p.s. I don't use cacheex, so users using it, please tell if something is broke with the patch.

comment:62 by theparasol, 11 years ago

Doesnt compile if cache-ex module is disable in make config
Add a #ifdef CS_CACHEEX in oscam-ecm.c lines 43~49 and oscam-work.c line 288

Last edited 11 years ago by theparasol (previous) (diff)

by blueven, 11 years ago

Attachment: ecm_handler_2.patch added

tryfix 2 for ecm handler. Tested on #8664

comment:63 by FilipeAmadeuO, 11 years ago

ecm_hanlder_2.patch build error on mca:

build/arm-none-linux-gnueabi-mca/module-cccam.o: In function `cc_parse_msg':
/home/filipe/Documents/simplebuild/oscam-svn/module-cccam.c:2713: undefined reference to `get_fallbacktimeout'
build/arm-none-linux-gnueabi-mca/oscam-ecm.o: In function `cw_process':
/home/filipe/Documents/simplebuild/oscam-svn/oscam-ecm.c:144: undefined reference to `get_fallbacktimeout'
collect2: ld returned 1 exit status
make[1]: * archiv/oscam-svn8664-mca-webif-patched.debug Error 1
make:
* [all] Error 2

by blueven, 11 years ago

Attachment: ecm_handler_4.patch added

ecm handler 4 (patch)

comment:64 by gf, 11 years ago

Don't mix white space changes with functional changes. If you clean your patch it'll be much more easier to see what is going on.

comment:65 by champagner, 11 years ago

All client losing its connection are disconnected from the server, why? The clients have been inactive for too long time, which is not true

The clients are disconnected from the server side constantly even though they are not client-side inactive and this behavior is only cccam2.1.4 and versions deeper.

2013/06/05 14:26:20 CC0C3C80 c [OSCAM-WORK] new event 1 occurred on fd 86 after 10594 ms inactivity
2013/06/05 14:26:20 CC0C3C80 c cccam(s) User_cccam2.1.4: disconnected by remote server
2013/06/05 14:26:20 CC0C3C80 c cccam(s) User_cccam2.1.4: connection closed by client, n=-1.
2013/06/05 14:26:20 CC0C3C80 c User_cccam2.1.4 disconnected from xxx.xxx.xxx.xxx
2013/06/05 14:26:20 CC0C3C80 c thread 7FA3D94CB700 ended!

2013/06/05 14:26:27 CC08C690 c [OSCAM-WORK] new event 1 occurred on fd 77 after 11940 ms inactivity
2013/06/05 14:26:27 CC08C690 c cccam(s) User2_cccam2.1.4: disconnected by remote server
2013/06/05 14:26:27 CC08C690 c cccam(s) User2_cccam2.1.4: connection closed by client, n=-1.
2013/06/05 14:26:27 CC08C690 c User2_cccam2.1.4 disconnected from xxx.xxx.xxx.xxx
2013/06/05 14:26:27 CC08C690 c thread 7FA3D954F700 ended!

There is something wrong

Blueven, please look in you PN Box at the streamboard!

comment:66 by FilipeAmadeuO, 11 years ago

ecm_handler_5 build error on MCA (matrix air cam)

CONF build/arm-none-linux-gnueabi-mca/config.c
LINK ../archiv/oscam-svn8687-mca-webif-dvbapi1v6-ecm_handler_5-patched.debug
build/arm-none-linux-gnueabi-mca/module-cccam.o: In function `cc_parse_msg':
/home/filipe/Documents/simplebuild/oscam-svn/module-cccam.c:2713: undefined reference to `get_fallbacktimeout'
build/arm-none-linux-gnueabi-mca/oscam-ecm.o: In function `cw_process':
/home/filipe/Documents/simplebuild/oscam-svn/oscam-ecm.c:165: undefined reference to `get_fallbacktimeout'
collect2: ld returned 1 exit status
make[1]: * archiv/oscam-svn8687-mca-webif-dvbapi1v6-ecm_handler_5-patched.debug Error 1
make:
* [all] Error 2

by blueven, 11 years ago

Attachment: ecm_handler_5.patch added

comment:67 by theparasol, 11 years ago

Simplebuild mca patches_on now builds correct a mca bin (ecm_handler_5.patch)

by blueven, 11 years ago

Attachment: ecm_handler_5-2.patch added

tryfix writelock/readlock problem.

by blueven, 11 years ago

Attachment: ecm_handler_5-3.patch added

fix cache stats

by blueven, 11 years ago

Attachment: ecm_handler_6.patch added

Make accessing ecm request by cacheex code "thread safe".

by blueven, 11 years ago

Attachment: ecm_handler_7.patch added

by blueven, 11 years ago

Attachment: ecm_handler_8.patch added

comment:68 by gf, 11 years ago

You are still mixing white space changes and functional changes. Stop doing that. Post diff using -w option or just clean your changes from the white space ones, leave only functional changes.

by blueven, 11 years ago

Attachment: ecm_handler_9.patch added

comment:69 by blueven, 11 years ago

-

Last edited 11 years ago by blueven (previous) (diff)

in reply to:  68 comment:70 by blueven, 11 years ago

Replying to gf:

You are still mixing white space changes and functional changes. Stop doing that. Post diff using -w option or just clean your changes from the white space ones, leave only functional changes.

Hi gf, I did a bit of cleaning. It's now ok?

comment:71 by chocholousek, 11 years ago

After r8680, if you change TV channel not change time on channel from web-if!

by blueven, 11 years ago

Attachment: ecm_handler_9-1.patch added

comment:72 by gf, 11 years ago

Much much better looking patch. I'm OK with it.

by blueven, 11 years ago

Attachment: ecm_handler_9-2.patch added

comment:73 by basta, 11 years ago

i cant see any disadvantage with patch 9.2, so why you dont put this patch into trunk?

by blueven, 11 years ago

Attachment: ecm_handler_10.patch added

Deatils on forum

by blueven, 11 years ago

Attachment: ecm_handler_10-1.patch added

fix bug from v10

by blueven, 11 years ago

Attachment: ecm_hanlder_11.patch added

Details on forum

by blueven, 11 years ago

Attachment: ecm_handler_11-1.patch added

by blueven, 11 years ago

Attachment: ecm_handler_11-2.patch added

by blueven, 11 years ago

Attachment: ecm_handler_11-3.2.patch added

by blueven, 11 years ago

Attachment: ecm_handler_11-3.patch added

comment:74 by blueven, 11 years ago

Not consider ecm_handler_11-3.2.patch​.. duplicated of ecm_handler_11-3.patch​.

by blueven, 11 years ago

Attachment: ecm_handler_11-4.patch added

by blueven, 11 years ago

Attachment: ecm_handler_11-5.patch added

by blueven, 11 years ago

Attachment: ecm_handler_11-6.patch added

by blueven, 11 years ago

fix for 11-6 patch

by blueven, 11 years ago

Attachment: ecm_handler_11-7.patch added

by blueven, 11 years ago

Attachment: ecm_handler_11-8.patch added

by blueven, 11 years ago

Attachment: ecm_handler_11-9.patch added

by blueven, 11 years ago

Attachment: ecm_handler_12.patch added

comment:75 by idioteque, 10 years ago

Is anyone else still getting the "No Matching Reader" error?

I'm using Oscam 9131 and getting "no matching reader" errors on channels I have local.

2013/12/21 11:16:19 B72C1458 c John1 (0963&000000/FFFF/183A/5F:C154328E4FCE94900D028F08C5FBFD3D): found (196 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:16:27 B72C1458 c John1 (0963&000000/FFFF/183A/5F:B010D2555F4126C857BFB5A8C37D9F95): rejected group (1 ms) - Animal Planet (no matching reader)
2013/12/21 11:16:27 B72C1458 c John1 (0963&000000/FFFF/183A/5F:B010D2555F4126C857BFB5A8C37D9F95): cache1 (0 ms) by 1 - Animal Planet
2013/12/21 11:16:35 B72C1458 c John1 (0963&000000/FFFF/183A/5F:DA9B9C6A4B6F5C5D886222C75C53607A): cache1 (60 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:16:43 B72C1458 c John1 (0963&000000/FFFF/183A/5F:7AFA3B39E105098F3FDF2DADFF1BA6FC): cache1 (87 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:16:51 B72C1458 c John1 (0963&000000/FFFF/183A/5F:95A113AC5A03B46C067ACF77D42A7C45): rejected group (0 ms) - Animal Planet (no matching reader)
2013/12/21 11:16:51 B72C1458 c John1 (0963&000000/FFFF/183A/5F:95A113AC5A03B46C067ACF77D42A7C45): cache1 (0 ms) by 1 - Animal Planet
2013/12/21 11:16:59 B72C1458 c John1 (0963&000000/FFFF/183A/5F:896A4CD33451F52F336C77DF0CD77936): rejected group (1 ms) - Animal Planet (no matching reader)
2013/12/21 11:17:07 B72C1458 c John1 (0963&000000/FFFF/183A/5F:1683D82618D02620820C0758572AB818): cache1 (30 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:17:15 B72C1458 c John1 (0963&000000/FFFF/183A/5F:DBE8A10117C0B042B48F3EED08D3776D): cache1 (56 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:17:23 B72C1458 c John1 (0963&000000/FFFF/183A/5F:D8AA675702635479BD1DF7C16A9328A5): rejected group (1 ms) - Animal Planet (no matching reader)
2013/12/21 11:17:23 B72C1458 c John1 (0963&000000/FFFF/183A/5F:D8AA675702635479BD1DF7C16A9328A5): rejected group (1 ms) - Animal Planet (no matching reader)
2013/12/21 11:17:23 B72C1458 c John1 (0963&000000/FFFF/183A/5F:D8AA675702635479BD1DF7C16A9328A5): cache1 (0 ms) by 1 - Animal Planet
2013/12/21 11:17:31 B72C1458 c John1 (0963&000000/FFFF/183A/5F:1FB8423C91EDD72E200F3028B59CF35D): rejected group (1 ms) - Animal Planet (no matching reader)
2013/12/21 11:17:31 B72C1458 c John1 (0963&000000/FFFF/183A/5F:1FB8423C91EDD72E200F3028B59CF35D): cache1 (0 ms) by 1 - Animal Planet
2013/12/21 11:17:39 B72C1458 c John1 (0963&000000/FFFF/183A/5F:98A90D8C9D1E07CD06486B85CF6E6467): cache1 (52 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:17:47 B72C1458 c John1 (0963&000000/FFFF/183A/5F:F8833E7FE9C58AC1B28F35CC1C46A989): cache1 (49 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:17:55 B72C1458 c John1 (0963&000000/FFFF/183A/5F:A4F24AADDE557DC16C9A5B8AF5C0AD0B): cache2 (56 ms) by 1 (L/3/3/5) - Animal Planet
2013/12/21 11:18:03 B72C1458 c John1 (0963&000000/FFFF/183A/5F:D170BFE4C81C67F8C95FB0782C1EBF75): rejected group (1 ms) - Animal Planet (no matching reader)
2013/12/21 11:18:03 B72C1458 c John1 (0963&000000/FFFF/183A/5F:D170BFE4C81C67F8C95FB0782C1EBF75): cache1 (0 ms) by 1 - Animal Planet
2013/12/21 11:18:11 B72C1458 c John1 (0963&000000/FFFF/183A/5F:424DB33362DCD064BF04744EDDA70422): cache2 (53 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:18:19 B72C1458 c John1 (0963&000000/FFFF/183A/5F:A825D8DED2BEE1997A0F29EB8EFF52E6): cache1 (57 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:18:27 B72C1458 c John1 (0963&000000/FFFF/183A/5F:68FFC25A3DC87BD395E7504029358A2A): cache1 (52 ms) by 1 (L/1/1/5) - Animal Planet
2013/12/21 11:18:35 B72C1458 c John1 (0963&000000/FFFF/183A/5F:2E6BD5EDB6EAC099ED0349BD2274444D): rejected group (1 ms) - Animal Planet (no matching reader)
2013/12/21 11:18:35 B72C1458 c John1 (0963&000000/FFFF/183A/5F:2E6BD5EDB6EAC099ED0349BD2274444D): cache1 (0 ms) by 1 - Animal Planet
2013/12/21 11:18:43 B72C1458 c John1 (0963&000000/FFFF/183A/5F:86861B245D0CE0FDF8373DC26976B34A): rejected group (0 ms) - Animal Planet (no matching reader)
2013/12/21 11:18:51 B72C1458 c John1 (0963&000000/FFFF/183A/5F:8D2BB04463A9BE98134269853D31BED5): cache2 (14 ms) by 1 (L/1/1/5) - Animal Planet

comment:76 by Deas, 9 years ago

Resolution: expired
Status: newclosed
Note: See TracTickets for help on using tickets.