Opened 14 years ago

Closed 14 years ago

#690 closed defect (fixed)

#2897

Reported by: pateras Owned by:
Priority: minor Component: General
Severity: medium Keywords:
Cc: Sensitive: no

Description

Hi to all, I have the OSCAM 0.99.4svn build #2897. After some minutes the oscam stop to work

Please find bellow the error messages

ERROR reading from fd 4 select_ret=-1, errno=4
PANIC: master deadlock! last location: event: udp-socket2010/08/04 9:38:03 18654 s cardserver down
ro

Ps: I m using smargo with version 1.30. Smargo work on oscam with smartreader mode

Thank you

Change History (12)

comment:1 by pateras, 14 years ago

This problem occured after 1 hour with and without cardreader. Im using ubuntu 10.04 with kernel Linux ubuntu64 2.6.32-24-server #38-Ubuntu SMP Mon Jul 5 10:29:32 UTC 2010 x86_64 GNU/Linux

Any solution?

Thanks

comment:2 by moicas3004, 14 years ago

Severity: mediumhigh

Im using also the latest oscam version for AZBOX and when working as server ( 30 readers and 30 users ) the oscam also stops working after a few minutes. As client only it works great.

comment:3 by alno, 14 years ago

Priority: criticalminor
Severity: highmedium

please post the content of /tmp/oscam.version

BTW: just because YOU have the problem the severity isn't high automatically. Normally we handle tickets which are tagged as high by single users with the lowest priority.

comment:4 by pateras, 14 years ago

I have the same problem with the latest svn build of oscam.
It generally stops working after 1 hour either in client mode or local sharing mode (with readers). I've also tried to change the platform on which the oscam is running. I've tried the PC Engines ALIX.1D with 256 MB RAM (x86_32) and Supermicro X7SPA with 4 Gig RAM (x86_64). I'm using Ubuntu Linux 10.04 on both platforms (either 32 bit version or 64 bit version). Please find bellow the content of oscam.version file.

uxstarttime: 1281278315
starttime: 08.08.10 17:38:35
version: 0.99.4svn#2902
maxpid: 512
webifsupport: yes
dvbapisupport: yes
gboxsupport: no
anticascsupport: yes

Finally, please note that I'm not facing any problems when running OSCAM on my Dreambox 7000 (PPC build). It seems that there is an issue on the x86 code of oscam.

Thanks for your support

comment:5 by manio, 14 years ago

Hello
I also have this error. I am using latest oscam (compiled yesterday from trunk) version 0.99.4svn, build #2902. After approx several minutes (yesterday it was 2,5min) i have the following error and oscam is shutting down (i was trying it also about 2-3 months ago and i've got same error then):

PANIC: master deadlock! last location: event: udp-socket

I think the problem is related with the situation, that half of clients are connecting with wrong DES and are rejected, then they are connecting again and so on (i have very much "Connection closed to client"). I am trying to replace two oscam instances (with different DES-es) into one with two cards (so that is the reason of half of clients connecting temporary with wrong DES). I know that this is wrong, BUT: it should not be a problem to oscam - it should reject the user, but not going into deadlock after some time of running.

comment:6 by alno, 14 years ago

please check with

[global]
resolvedelay=0

should be similar as described here: http://www.www.streamboard.tv/wbb2/thread.php?postid=343275#post343275

comment:7 by manio, 14 years ago

Cc: manio@… added

@alno
thank you for your tip, but unfortunatelly it doesn't change anything :(
after 9 minutes of running i've got:

s   ERROR reading from fd 3 select_ret=-1, errno=4

then all clients get's emms only from cache for a short while and after several seconds oscam finally stopped like usual:

PANIC: master deadlock! last location: event: udp-socket
s   cardserver down

please notice that separate oscam instances works for me without problems, but i'd like to have one oscam instance with two cards

ps. just by the way: please apply below patch (cosmetic fix):

Index: oscam.c
===================================================================
--- oscam.c     (revision 2902)
+++ oscam.c     (working copy)
@@ -288,7 +288,7 @@
 static void cs_master_alarm()
 {
   cs_log("PANIC: master deadlock! last location: %s", mloc);
-  fprintf(stderr, "PANIC: master deadlock! last location: %s", mloc);
+  fprintf(stderr, "PANIC: master deadlock! last location: %s\n", mloc);
   fflush(stderr);
   cs_exit(0);
 }

comment:8 by pateras, 14 years ago

Any solution about this problem?

Thanks for your support

comment:9 by manio, 14 years ago

Cc: manio@… removed

comment:10 by manio, 14 years ago

Hi guys.
I am not sure about that - but i think it could be a workaround:
add two options to global section of oscam.conf:

clienttimeout = 7
resolvedelay = 0

it works me longer than usual on theese settings, but it shutdown during the night, unfortunatelly don't know when and what was the reason. Please try it and let know about result - it could be valuable information for developers ... :)
regards

comment:11 by DrEvil, 14 years ago

It happens to me when clients disconnect it seems. I will try to get log info when it next happens, but this is just my observation. Cheers, Evil

comment:12 by Deas, 14 years ago

Resolution: fixed
Status: newclosed

there has been a big change in the ocam architecture now. so please check again with the latest threaded version and reopen a new ticket with debug log if problem still exists.

Note: See TracTickets for help on using tickets.