Hi, I have a trivial setup on a 2.4GHz Xeon Dell PE 1750 SCSI machine dealing with 4 ports of E1 in an 'inline PBX' arrangement. My extensions.conf is simply: [general] static=yes writeprotect=yes [frompstn] exten => _31.,1,Dial(Zap/g2/${EXTEN}) exten => _31.,2,Congestion [fromaxxess] exten => _13.,1,Dial(SIP/${EXTEN},,h) exten => _13.,2,Congestion exten => _31.,1,Dial(Zap/g2/${EXTEN}) exten => _31.,2,Congestion include => outbound [outbound] exten => _X.,1,Dial(Zap/g1/${EXTEN}) exten => _X.,2,Congestion We have a full 30-channel PRI and a 4-channel partial PRI and are experiencing load spikes that I can't find the source of. The machine Debian sarge on the default 2.6.8-2-686 kernel, and no other daemons are running than sshd. The machine is doing no IP work - purely TDM, yet on a Xeon 2.4GHz machine, the load average is sitting at 0.6 with 40 active Zap channels (i.e. 20 live calls) and will randomly jump to 2 (with call quality starting to stutter) A few seconds of vmstat: procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 7 0 0 223560 1276 223040 0 0 3 10 83 94 1 2 97 0 0 0 0 223552 1284 223040 0 0 0 16 5128 3461 1 0 98 1 0 0 0 223552 1284 223040 0 0 0 0 5094 3319 10 9 81 0 0 0 0 223552 1284 223040 0 0 0 16 5130 2955 1 10 89 0 0 0 0 223552 1292 223040 0 0 0 60 5121 2918 0 1 97 2 0 0 0 223552 1292 223040 0 0 0 0 5031 2936 1 0 99 0 Does this sound about normal for what is just shuffling data between ports of the Sangoma A104? I want to record the call data with the 'Monitor' application but this just causes the load to increase even more (even though 'hdparm' shows 70MB/sec disk transfer with low user+system CPU usage) /proc/interrupts is CPU0 0: 423253622 IO-APIC-edge timer 1: 175 IO-APIC-edge i8042 9: 0 IO-APIC-level acpi 11: 0 IO-APIC-level ohci_hcd 12: 58 IO-APIC-edge i8042 15: 13 IO-APIC-edge ide1 177: 533330 IO-APIC-level ioc0 185: 29 IO-APIC-level ioc1 193: 1311243931 IO-APIC-level wanpipe1, wanpipe2, wanpipe3, wanpipe4 201: 13289965 IO-APIC-level eth0 217: 5420038 IO-APIC-level eth2 NMI: 0 LOC: 423311408 ERR: 0 MIS: 0 Help! :) Cheers, Gavin.
Mojo with Horan & Company, LLC
2005-Nov-29 10:35 UTC
[Asterisk-Users] Load spikes with 1.0.10
Are your interrupts getting hogged by anything else? I'd recommend http://www.voip-info.org/wiki/view/Asterisk+PCI+bus+Troubleshooting if you haven't already read it. Have you tried booting with noapic kernel option? You may then have to shuffle cards around to make your sangoma not share any interrupts <shrug> hth :) moj Gavin Hamill wrote:> Hi, I have a trivial setup on a 2.4GHz Xeon Dell PE 1750 SCSI machine > dealing with 4 ports of E1 in an 'inline PBX' arrangement. > > My extensions.conf is simply: > > [general] > static=yes > writeprotect=yes > > [frompstn] > exten => _31.,1,Dial(Zap/g2/${EXTEN}) > exten => _31.,2,Congestion > > [fromaxxess] > exten => _13.,1,Dial(SIP/${EXTEN},,h) > exten => _13.,2,Congestion > exten => _31.,1,Dial(Zap/g2/${EXTEN}) > exten => _31.,2,Congestion > include => outbound > > [outbound] > exten => _X.,1,Dial(Zap/g1/${EXTEN}) > exten => _X.,2,Congestion > > We have a full 30-channel PRI and a 4-channel partial PRI and are > experiencing load spikes that I can't find the source of. > > The machine Debian sarge on the default 2.6.8-2-686 kernel, and no other > daemons are running than sshd. > > The machine is doing no IP work - purely TDM, yet on a Xeon 2.4GHz > machine, the load average is sitting at 0.6 with 40 active Zap channels > (i.e. 20 live calls) and will randomly jump to 2 (with call quality > starting to stutter) > > A few seconds of vmstat: > > > procs -----------memory---------- ---swap-- -----io---- --system-- > ----cpu---- > r b swpd free buff cache si so bi bo in cs us sy > id wa > 7 0 0 223560 1276 223040 0 0 3 10 83 94 1 2 > 97 0 > 0 0 0 223552 1284 223040 0 0 0 16 5128 3461 1 0 > 98 1 > 0 0 0 223552 1284 223040 0 0 0 0 5094 3319 10 9 > 81 0 > 0 0 0 223552 1284 223040 0 0 0 16 5130 2955 1 10 > 89 0 > 0 0 0 223552 1292 223040 0 0 0 60 5121 2918 0 1 > 97 2 > 0 0 0 223552 1292 223040 0 0 0 0 5031 2936 1 0 > 99 0 > > Does this sound about normal for what is just shuffling data between > ports of the Sangoma A104? I want to record the call data with the > 'Monitor' application but this just causes the load to increase even > more (even though 'hdparm' shows 70MB/sec disk transfer with low > user+system CPU usage) > > /proc/interrupts is > CPU0 > 0: 423253622 IO-APIC-edge timer > 1: 175 IO-APIC-edge i8042 > 9: 0 IO-APIC-level acpi > 11: 0 IO-APIC-level ohci_hcd > 12: 58 IO-APIC-edge i8042 > 15: 13 IO-APIC-edge ide1 > 177: 533330 IO-APIC-level ioc0 > 185: 29 IO-APIC-level ioc1 > 193: 1311243931 IO-APIC-level wanpipe1, wanpipe2, wanpipe3, wanpipe4 > 201: 13289965 IO-APIC-level eth0 > 217: 5420038 IO-APIC-level eth2 > NMI: 0 > LOC: 423311408 > ERR: 0 > MIS: 0 > > Help! :) > > Cheers, > Gavin. > > _______________________________________________ > --Bandwidth and Colocation provided by Easynews.com -- > > Asterisk-Users mailing list > To UNSUBSCRIBE or update options visit: > http://lists.digium.com/mailman/listinfo/asterisk-users >-- Mojo <mojo@horanappraisals.com> Office Manger, Horan & Company, LLC (907) 747-6666 x112