![]() | Oracle System Handbook - ISO 7.0 May 2018 Internal/Partner Edition | ||
|
|
![]() |
||||||||||||||||||
Solution Type Problem Resolution Sure Solution 2094270.1 : When comm-monitor (SBC-as-a-Probe) is Using a Media Interface, During Peak Time Some of the Traffic is Dropped/not Sent to OCSM/EOM
In this Document
Applies to:Oracle Communications Session Monitor - Version 3.1.x and laterAcme Packet 6300 - Version S-Cz7.2.0 and later Net-Net OS - Version S-Cx6.3.0 and later Information in this document applies to any platform. SymptomsWhen comm-monitor (SBC-as-a-probe) is set to send data through a media interface, during peak time some of the traffic is dropped/not sent to OCSM/EOM The "show acl summary" command on the SBC shows dropped untrusted packets CauseOn media interfaces, IPFIX traffic (SBC-as-a-probe to OCSM/EOM) is considered untrusted, thus the SBC is limiting the bandwidth and, as a consequence, dropping messages. This does not apply to the management/wancom0 interface. SolutionThere are 3 possible options that can be explored:
1. Increase max-untrusted-signaling and/or max-signaling-bandwidth and check performance in a controlled way Increase the max-untrusted-signaling parameter under media-manager. Its value is a percentage of the max-signaling-bandwidth, also defined in media-manager There is no recommended value to set these parameters to, thus we recommend to gradually increase the max-untrusted-signaling parameter under media-manager (and/or max-signaling-bandwidth), each time monitoring the behavior of the SBC and OCSM/OEM ME. The idea behind this change is to increase the amount of bandwidth (internal bandwidth in the SBCs) that can be used on the SBC by untrusted traffic, taking into account that IPFIX traffic is considered untrusted. Please find below the step by step procedure: # conf t (configure)# media-manager (media-manager)# media-manager (media-manager-config)# sel (media-manager-config)# max-untrusted-signaling <chosen value> (media-manager-config)# done … (media-manager-config)# exit (media-manager)# exit (configure)# exit # save-config # activate-config This change does not require a reboot and it can be done anytime (no service affecting). Since the parameter should be fine tuned gradually until we find the best value, there is only need to perform this change in 1 SBC pair, then monitor it for several hours. Ideally would be to monitor the behavior of the SBC after the change during the peak hours, or preferably 24 hours period (so we have the complete picture). What needs to be monitored is the amount of dropped messages in the trusted and untrusted queues, which can be done with the "show acl summary" command (last column on the right). Here is a sample output: sbc# show acl summary
Entries Packets Dropped Total deny entries: 2 N/A 0 Total media entries: 0 Total trusted entries: 25892 2831918 0 Total untrusted entries: 6 754859 0 Totals by interface: Slot 0/Port 0: Deny entries: 1 N/A 0 Untrusted entries: 2 153134 0 Trusted entries: 25890 2831918 0 Slot 1/Port 0: Deny entries: 1 N/A 0 Untrusted entries: 4 601725 0 Trusted entries: 2 0 0 Ideally, the number of dropped messages should be 0 or just a small number; what we recommend to check is: - The number of dropped messages in the untrusted entries should be reduced. - The number of dropped messages in the trusted entries should not be increased. This command can be executed regularly to see how it behaves.
2. Move all IPFIX traffic to the management interface. This issue not apply to the management/wancom0 interface, so you could change the configuration the comm-monitor (SBC-as-a-probe) connection to use wancom0 instead of media interface. However, please note that the media interface is faster in sending messages that wancom0. This means that in cases of high traffic, using wancom0 can sometimes lead to packets getting dropped internally on the SBC. If you move to the wancom0 interface, you should monitor the 'Socket Message Dropped' counter from the "show comm-monitor stats" output, here is an example: # show comm-monitor stats
Client stats Client 10.1.29.26:4739 (CONNECTED) Stats Communication Monitor Stats ---- Lifetime ---- Recent Total PerMax Handshake Msg Sent 0 61 2 Handshake Msg ACK 0 61 2 Handshake Msg NAK 0 0 0 Keep Alive 0 0 0 SIP UDP Recv Msg Sent 115460 12154518 138106 SIP UDP Send Msg Sent 114237 12028456 136170 SIP TCP Recv Msg Sent 0 0 0 SIP TCP Send Msg Sent 0 0 0 ... QoS Data Sent 11421 1177892 13002 Socket Message Sent 241118 25360927 287278 Socket Message Dropped 0 17437 1401 <------------------- If this counter increases then it means messages are dropped Socket Send Error 0 17437 1401 Socket Not Ready 0 0 0 ... See also Bug 25542810 - Broken IPFIX messages If you do see "Socket Message Dropped" counter increasing, see also: Missing Data and Intermittent IPFIX Errors Seen: "ipfix msg version" for SBC Probes (<Document 2285038.1>)
3. Request assistance from Oracle Professional Services (through your Sales Contact or Account Manager) Our Professional Services team can help analyze the traffic patterns and configuration needs and assist customers in reaching the best configuration for their SBC. Attachments This solution has no attachment |
||||||||||||||||||
|