Port Allocation failed

kelvintee · Post by **kelvintee** » Sun Apr 22, 2012 12:04 am

1. Brekeke Product Name and version:
Brekeke SIP Server v.3 Advanced edition

2. Java version:
1.6.0_31

3. OS type and the version:
CentOS 4.6

4. UA (phone), gateway or other hardware/software involved:
Class 4 softswitch

5. Select your network pattern from http://www.brekeke-sip.com/bbs/network/ ... terns.html :
9

6. Your problem:
Received Port allocation failed in call logs during busy hour.

RTP Configuraiton as below:
RTP relay = ON
RTP Relay (UA) = Off
Min Port: 10000
Max Port: 29999
RTP sessions: 5000
Port Mapping: sdp
RTP session timeout = 600000

Centos limits.conf = soft 32768, hard 63536

Busy hour traffic session in peak = 350 - 370 concurrent session

Server hardware spec:
CPU: Intel Xeon QUAD core 2.5Ghz
Memory: 8GB

I have follow exact wiki brekeke instruction and adjust configuration accordingly but still receiving the error message. Please advise which part I missed out. Thanks

hope · Post by **hope** » Mon Apr 23, 2012 9:55 am

is there any error message in sv log?
sv log is at webapps\sip\WEB-INF\work\sv\log\2012\04\

redroof · Post by **redroof** » Mon Apr 23, 2012 10:41 am

What kind of SIP UA are you using?

How about memory usage?
Are there enough free memory during busy hour?

kelvintee · Post by **kelvintee** » Tue Apr 24, 2012 3:03 am

hope wrote:is there any error message in sv log?
sv log is at webapps\sip\WEB-INF\work\sv\log\2012\04\

Yes. Following keep repeating appear in sv.log
Exception: 04/23/12 21:28:21.957
java.lang.NullPointerException
com.brekeke.net.sip.sv.SIPex.runImple (null:-1)
com.brekeke.thread.ThreadCommon.run (null:-1)

and Proxy.log having following error message:

************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999
************************* port open all failed min:10000 max:29999

kelvintee · Post by **kelvintee** » Tue Apr 24, 2012 3:05 am

redroof wrote:What kind of SIP UA are you using?

How about memory usage?
Are there enough free memory during busy hour?

I don't used SIP UA. Those traffic direct terminate from softswitch, I used the SIP server act as SIP proxy server for topology hiding. Porxy SIP signalling and RTP path.

No memory issue as during incident period hardware usage pretty low.

Cpu(s): 0.7% us, 5.4% sy, 0.0% ni, 93.5% id, 0.2% wa, 0.2% hi, 0.0% si
Mem: 8306028k total, 665236k used, 7640792k free, 115672k buffers
Swap: 2096472k total, 0k used, 2096472k free, 257428k cached

redroof · Post by **redroof** » Tue Apr 24, 2012 9:28 am

try the following.

Add net.rtp.native=false in the [Configuration]->[Advanced] page, and restart the SIP Server.

kelvintee · Post by **kelvintee** » Tue Apr 24, 2012 9:09 pm

redroof wrote:try the following.

Add net.rtp.native=false in the [Configuration]->[Advanced] page, and restart the SIP Server.

Hi Redroof,

Thank you for the suggestion. May I know what is the impact on above changes? Will it affect my proxy RTP & Signalling? I couldn't find any info on this particular setting in wiki brekeke. Thanks

Regards,
Kelvin

kelvintee · Post by **kelvintee** » Wed Apr 25, 2012 12:22 pm

redroof wrote:try the following.

Add net.rtp.native=false in the [Configuration]->[Advanced] page, and restart the SIP Server.

Hi Redroof,

I tried but still getting port allocation failed message in session log during busy hour. What else I can try?

redroof · Post by **redroof** » Wed Apr 25, 2012 6:02 pm

in my environment, there are more than 700 concurrent calls. the peak cps is around 120.

I doubt a port confliction problem happened in your machine.
Let you check it.

Make sure close other applications which may use the port 10000 -29999.
Open these ports on the firewall.

kelvintee · Post by **kelvintee** » Thu Apr 26, 2012 12:57 am

redroof wrote:in my environment, there are more than 700 concurrent calls. the peak cps is around 120.

I doubt a port confliction problem happened in your machine.
Let you check it.

Make sure close other applications which may use the port 10000 -29999.
Open these ports on the firewall.

Port open failed error message only happen once which on 21 Apr 2012 . Believe could be due to Am shutdown & start SIP server too quickly causing port conflict. for the last few days, no more such error message.

I still getting "Port Allocation failed" during busy hour. Can you share with me which edition you used? and Your Proxy server configuration setting. Mine as below:

#SIP Server properties file
#Thu Apr 26 13:10:01 SGT 2012
net.registrar.upper.url=
net.usrdir.plugins=SqlUserDir
net.rtp.video.port.max=0
net.rtp.loglevel.console=0
alias.url=
net.upnp.use=false
net.auth.realm=proxy1.ngtnetworks.com
net.net1.interface=203.208.204.3
userdir.url=
sdpadmin.ver=2
tmp.radius.acctport=1813
sipadmin.net.upnp.timer.period.sec=30
net.sip.b2bua=false
net.sip.transport.try.udp=true
net.registrar.thru.allow=on
net.rtp.port.min=20000
alias.enabled=off
net.registrar.loglevel.console=0
sipadmin.productname=
ua.loglevel.console=0
net.rtcp.loglevel.console=0
net.registrar.thru.nat.allow=true
net.ipv6.policy.default=true
net.sip.tcp.use=false
net.register.loglevel.console=0
sipadmin.net.portmap.cache.period.sec=86400
net.sip.tcp.size.queue=50
path.work=/opt/tomcat/webapps/proxy/WEB-INF/work/sv
networkaddress.cache.ttl=3600
net.net4.interface=
sv.location=Equinix DC
net.sip.key.ignoreuri=false
com.brekeke.sipadmin.startparam1=-Xms1024m -Xmx2048m
net.sip.adjustif=off
net.registrar.upper.catchhere=true
tmp.net.registrar.upper.protocol=UDP
net.sip.ipfilter.pattern=
alias.user=
userdir.driver=
tmp.radius.acct.send.stop.onfailure=false
net.sip.port.local=5060
net.auth.policy4=off
sipadmin.startup=auto
net.auth.policy3=off
net.auth.policy2=off
net.sip.res100.all=off
net.portmap.cache.size=24
net.rtp.port.max=29999
net.userdir.cryptpw=true
net.sip.addrport=on
userdir.enabled=off
net.sipex.precreate=10
keyfile.cert.der=
keyfile.key.der=
alias.driver=
net.net2.interface=
net.sip.followroute.nat=false
net.auth.request_r=off
sv.description=NGTNetworks SIP proxy Server
net.auth.loglevel.console=0
sipadmin.tls.keyfile.type=DER
net.auth.request_i=on
keyfile.keystore.jks=
net.sip.timeout.talking=3600000
net.listener.loglevel.console=0
net.ip-gw=
net.sip.addrport.recv=off
net.registerex.precreate=0
net.net5.interface=
net.sip.tls.size.queue=50
net.registrar.thru.catchhere=true
net.auth.nonce-expires=60
net.userdir.loglevel.console=0
net.dns.aaaa=true
registered.user=
tmp.radius.radiushost=203.208.204.6
net.ipv6.use=false
embedded.port=
net.dns.srv=true
userdir.user=
sv.admin.sip=proxy1.ngtnetworks.com
net.external-addr.pattern=
net.sip.timeout.ringing=60000
sv.name=proxy1-ngtnetworks
net.rtp.session.timeout=600000
tmp.radius.sharedsecret=brekeksip
net.sip.tls.accept.anycert=false
net.registrar.upper.nat.allow=true
net.rtp.responseport=off
net.registrar.awake.use=off
net.rtp.relay.localhost=off
net.sip.ipfilter.type=disable
registered.driver=
net.rtp.video.port.min=0
registered.enabled=off
net.registrar.maxsession=100
net.registrar.upper.timeout=30000
net.registrar.awake.interval=12000
mng.loglevel.console=0
tmp.radius.authport=1812
net.sipex.session.max=10
sv.loglevel.console=0
sv.admin.mail=
tmp.radius.onoff=off
net.sip.loglevel.console=0
net.sip.tls.use=false
net.net3.interface=
net.session.limit=-1
net.upnp.port.exact=false
net.registerex.job.max=10
net.registrar.upper.allow=off
registered.url=
net.rtp.relay=on

redroof · Post by **redroof** » Tue May 01, 2012 11:30 am

I didn't set any special settings. It means almost default.
I set the RTP-relay=on.

Have you tried to use another server hardware?

kelvintee · Post by **kelvintee** » Wed May 02, 2012 7:32 am

redroof wrote:I didn't set any special settings. It means almost default.
I set the RTP-relay=on.

Have you tried to use another server hardware?

My Server is Quad Core 8GB ram so hardware should be good enough. I don't have other server hardware to try.

I am running on Centos v4.6. Please share with me what OS you used? thanks

redroof · Post by **redroof** » Thu May 03, 2012 11:49 am

Im using several Windows including XP and 7.

redroof · Post by **redroof** » Thu May 03, 2012 11:51 am

Did you execute "netstat -a" command to see how UDP ports are consumed?

kelvintee · Post by **kelvintee** » Fri May 04, 2012 7:37 am

redroof wrote:Did you execute "netstat -a" command to see how UDP ports are consumed?

Yes, everything seem normal. infect using java tools to manual open 2000 UDP port and it can open all port without any problem.

kelvintee · Post by **kelvintee** » Fri May 04, 2012 7:38 am

redroof wrote:Im using several Windows including XP and 7.

Wonder will it CentOS 4.6 causing this issue.

redroof · Post by **redroof** » Fri May 04, 2012 10:54 am

Can you upgrade the CentOS to the latest version?

kelvintee · Post by **kelvintee** » Sun May 06, 2012 8:41 pm

redroof wrote:Can you upgrade the CentOS to the latest version?

Yes. going to run brekeke SIP server on latest CentOS 5.8 version.

kelvintee · Post by **kelvintee** » Mon May 14, 2012 1:58 am

kelvintee wrote:
redroof wrote:Can you upgrade the CentOS to the latest version?
Yes. going to run brekeke SIP server on latest CentOS 5.8 version.

Everything working fine in CentOS 5.8. No cruel why CentOS 4.9 got got such problem.

redroof · Post by **redroof** » Mon May 14, 2012 11:18 am

I'm not sure but it seems the issue was caused by CentOS 4's bug.

Anyway I'm glad to know you made it work.