openUC Release 14.04.4
Several bug fixes that have been in the staged build for 14.04.3 for a while are being wrapped into the 14.04.4 release. We decided to make this release available for those not ready to move to 14.10 yet (see separate announcement of 14.10.1).
In optimizing performance issues in release 14.04.3 some log file changes caused an over abundance of logging and log files to not get rotated properly, this can lead to disk space getting consumed rapidly. This has been corrected in 14.04.4.
Also in 14.04.3 we had focused on anything that could affect performance of the proxy or registrar. Freeswitch’s execution priority was dropped to prevent possible CPU starvation. This fix has reverted because in certain installations this caused static with MoH and poor quality conference bridge calls. This was seen mostly in larger virtualized environments.
In highly available systems slight delays in database replication would cause registrations to drop for a short period of time. A grace period for registrations has been implemented to account for any delays that may occur during replication.
An issue with TCP signalling and sipXproxy was addressed that caused some TCP packets to be missed was addressed.
A new proxy plugin was added to allow openUC to allow better interoperation with Oracle’s Acme Packet SBC’s.
Three Reach related issues are also addressed in 14.04.4. Agents pressing Do Not Disturb on their phones caused an issue that had the system continuously re-offering their station the call. Another problem that appeared randomly, without any given pattern, was causing a communications error inside freeswitch. And inbound calls that had a single quote symbol in the caller-id would cause agents to go offline.
- Full Release Notes are located here: http://www.ezuce.com/group/ezuce-documentation/documentation/-/wiki/eZuceTechnicalDocumentation/EZuce+openUC+14.04
Who Should Install?
This is a bug fix release that is recommended for all 14.04 installations.
If you have a patch installed to your system a new patch may be required. Please contact firstname.lastname@example.org if think you may have a patch applied as that may be replaced during the update.
If you have questions about updating you can email email@example.com or if you need assistance with the update contact your account manager or email firstname.lastname@example.org.
Specific Issues Addressed
|JIRA ID||JIRA name||RN Content|
|UC-3438||Create Proxy plugin to solve MoH / Ringing overlap for poorly behaving SBCs||The plugin will do the following1. Check if the sipfrag of a notify is a 180 Ringing2. Convert to sipfrag 100 Trying before sending to PolycomEnd result: Polycom will only send BYE upon receipt of sipfrag 200 Ok. The intention is to let MoH play the entire alerting phase.In the old use case where call goes to a mobile phone, sipfrag will be 183 Session Progress. This will not be changed by the plugin. This would result to MoH stopping to avoid the media overlap.1. Disabled by default
2. Will affect only notifies for event refer
3. Will not have any effect on routing, but would effect when polycom tears down MoH.
|XX-11584||Write race on the same TCP socket descriptor||Fixed an issue whereby the socket::send operations are not thread safe and there is the possibility of missing some TCP packets if two threads try to grab the same socket at the same time.The race condition applies only to TCP and the involved functions are SipClient::isWritable() and SipClientWriteBuffer::writeMore() since they are run in distinct threads.|
|XX-11462||Fix log rotate behavior where log file gets recreated every 5 seconds||Fixed in issue introduced in 14.04.3 where log file changes caused an over abundance of logging and log files to not get rotated properly, this can lead to disk space getting consumed rapidly.|
|XX-11551||Implement expire grace period for registrations||Fixed an in with that appears in distributed highly available systems where slight delays in database replication would cause registrations to drop for a short period of time. A grace period for registrations has been implemented to account for any delays that may occur during replication.|
|XX-11399*Revert*||Start FS process with normal priority||Reverted a fix introduced in 14.04.3. Freeswitch starts with higher priority than proxy, it was observed during load testing AAs that FS got a big slice of CPU and proxy didn’t have enough resources to process calls. Load testing AAs with 30cps went just fine after starting FS with lower priority (that is with -np option).In live environments with customers running certain virtualization technology, this change caused static in MoH and poor quality conference bridging.|
|UC-3274||Reach – Freeswitch communication error – freeswitch_ring:init bad uuid||Fixed an issue that appeared randomly, without any given pattern, was causing a communications error inside freeswitch.|
|UC-3405||Agent is put in released state if receives call from client with username containing ‘ char||Fixed an issue that caused agents to be placed in released state if the agent receives a call from a client queue that had a single quote in the caller-id.|
|UC-3508||When the agents phone is DND a call is offered over-and-over||Fixed an issue with Reach that was caused by Agents putting their phone on DND.Assume we have a Reach agent called John.John sets his Reach status as `Available`.After a while he puts his deskphone on DND.If a call is routed by Reach to this agent while in DND the call is offered again and again to this user, even if the phone returns 486 BUSY.|