See archives on the general uselessness of RH. Your fist try should be to recompile the kernels if for nothing else than the fact that once again there is 2 bug reports against the RH shipping kernels this week, and recently announced rootable bug in the shipping kernels. RH does not have a good track record as far as I can see for including their pet patches.
See if when you get a 2.4.20 custom built kernel if you don't get better performance. On Wed, 2003-06-11 at 07:03, Carlos Car�s wrote: > Hi! > > I have the chance to play with a couple of E400P cards, each installed > in a IBM e330 XSeries servers (2 x 1GHz P-III CPU 2 Gb RAM, 36Gb SCSI > HDD with RH8.0 2.4.18-smp kernel), and I'm trying to test/benchmark this > e330/E400P combo generating calls thru /var/spool/asterisk/outgoing > > One e400P if doing the carrier work making calls and the other just > receives the calls: > > Server#1 Server#2 > caller callee > +-------+ +-------+ > | Span1 |--E1 crossover cable-->| Span1 | > | Span2 |--E1 crossover cable-->| Span2 | > | Span3 |--E1 crossover cable-->| Span3 | > | Span4 |--E1 crossover cable-->| Span4 | > +-------+ +-------+ > > The basic configuration seems ok, since zttool shows the links are OK. > > I'm using this UTP cat5 cross-cable (not shielded): > pin1 <--> pin4 > pin2 <--> pin5 > pin3 <--> pin6 > pin4 <--> pin1 > pin5 <--> pin2 > pin6 <--> pin3 > pin5 <--> pin8 > pin8 <--> pin7 > > > I'm not really sure this is correct, since I've found only how to > connect pins 1,2,4 and 5. The other pins are connected as I supposed it > should be. > > This is my zapata.conf: > > Server#1 zapata.conf (Server#2 has the same zapata.conf but pri_net is > pri_cpe) > > -------------------------------------- > [channels] > context=inicio > switchtype=euroisdn > signalling=pri_net > rxwink=300 ; Atlas seems to use long (250ms) winks > usecallerid=yes > hidecallerid=no > callwaiting=yes > callwaitingcallerid=yes > threewaycalling=no > transfer=no > cancallforward=no > callreturn=no > echocancel=yes > echocancelwhenbridged=yes > rxgain=0.0 > txgain=0.0 > > group=1 > signalling=pri_net > channel => 1-15 > channel => 17-31 > channel => 32-46 > channel => 48-62 > channel => 63-77 > channel => 79-93 > channel => 94-108 > channel => 110-124 > -------------------------------------- > > To generate calls, I've done a small C proggie that generates files in > /var/spool/asterisk/outgoing/ like this one (filename and callerid are > different for each call): > > -------------------------------------- > Channel: Zap/g1 > Context: default > Extension: s > Priority: 1 > Callerid: 55512345 > -------------------------------------- > > The Dialplans are simple...caller machine just plays a 3 minutes gsm and > loops, and the callee machine dilaplan launches an AGI that plays some > gsm, records 20 secs of the call and hangups the call. > > Server#1 (caller) extensions.conf > > -------------------------------------- > [general] > static=yes > writeprotect=no > [inicio] > exten => s,1,PlayBack(laxana) > exten => s,2,Goto(s,1) > exten => t,1,hangup > exten => i,1,hangup > exten => o,1,hangup > exten => h,1,hangup > -------------------------------------- > > > Server#2 (callee) extensions.conf > -------------------------------------- > [general] > static=yes > writeprotect=no > [inicio] > exten => s,1,Answer > exten => s,2,Agi,600agi > exten => s,3,hangup > exten => t,1,hangup > exten => i,1,hangup > exten => o,1,hangup > exten => h,1,hangup > -------------------------------------- > > > Everything works pretty well with just 1 call, but my goal is to > generate a much high number (a queue of +50.000 calls), but I've run > into some problems... > > If I try to generate 120 simultaneous calls (top of lines available with > my 4 E1), I get a lot of errors in the logs. > I've tried to sleep(1) the begining of each call, and now I can only > generate a maximum of 60~80 simultaneous calls. Could it be due to a > cross-over cable problem or maybe the server can't deal with it? > > Btw, asterisk gets 100% of all available CPU (user CPU) for 10~20 > seconds and then it keeps about 80% CPU usage.... > > Thas's just a portion of the asterisk log: > -------------------------------------- > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: !! Got reject for frame 67, retransmitting frame 67 > now, updating n_r! > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: !! Got reject for frame 67, retransmitting frame 68 > now, updating n_r! > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: !! Got reject for frame 67, retransmitting frame 69 > now, updating n_r! > (...) > Jun 11 13:12:16 WARNING[90124]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: Short write: -1/16 (Unknown error 500) > Jun 11 13:12:16 WARNING[90124]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: Short write: -1/20 (Unknown error 500) > Jun 11 13:12:16 WARNING[90124]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: Read on 138 failed: Unknown error 500 > (...) > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: !! Got reject for frame 67, retransmitting frame 70 > now, updating n_r! > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: Short write: -1/20 (Unknown error 500) > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: !! Got reject for frame 67, retransmitting frame 71 > now, updating n_r! > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: Short write: -1/16 (Unknown error 500) > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: !! Got reject for frame 67, retransmitting frame 72 > now, updating n_r! > Jun 11 13:12:16 WARNING[81931]: File chan_zap.c, Line 5341 > (zt_pri_error): PRI: Short write: -1/20 (Unknown error 500) > (...) > -------------------------------------- > > > > Any suggestion, ideas, help very much appreciated. > > > Best Regards!! -- Steven Critchfield <[EMAIL PROTECTED]> _______________________________________________ Asterisk-Users mailing list [EMAIL PROTECTED] http://lists.digium.com/mailman/listinfo/asterisk-users
