Response time delay and log research

[Rich]

One of our clients submitted a batch of 400 calls today (7/11). Actually each call is submitted individually so that we can have better control over the retry process. A limit of 25 concurrent calls are submitted at a time to prevent too many calls from being concurrently processed and overwhelming either the Plum system or our web service. The client reported that they received over 100 calls from their callees that there was long delays between the time they answered the call and the time they heard anything, or hung up because the system did not say anything. Unfortunately they did not collect the phone numbers or other identifying information so I’m unable to research a specific call. I did a spot check of call logs that we keep and that I could find in Plum’s logs without any success on finding evidence of the issue. The majority of calls were made between 11am and 12 Eastern time, some before and more afterward as well.

So my first question looking for help on researching this kind of issue. The first issue I run into is that the detailed logs only support the last 100 calls on a given account. Of course that limits the number of call records that I can research. This will get even worse as we bring more clients with larger call volumes on line. Any suggestions on research would be helpful.

Second, the detailed call log does not seem to have the level of detail needed to track time delays. In fact the limited time stamps in it seem a little confusing as it indicates 3 voice prompts are prompts are played and then a second later 2 more voice prompts are played. Clearly all three voice prompts could not be played in a single second so I’m not sure what the call log represents. Maybe this is just a lack of understanding how to read the log.

Third, in the log what would really help is knowing what time the fetch of VoiceXML was call and returned. This way if the delay is a result of a slow response by our servers we could tell this clearly in the log.

Fourth, maybe a long shot question but do you know if the Plum Voice systems were experiencing any system issues around this time that could have caused they delays?

Thanks for any feedback and help you can offer,

Rich

[support]

Hi Rich,

Regarding issue #1 with only the last 100 call logs being available for viewing, it is possible for us to increase this limit for your account to look at more logs throughout the day. However, for this request, you would need to contact support@plumgroup.com to increase the limit on call logs and there is currently an upper limit of 750 total call logs.

Regarding issue #2 about looking through the call logs, the timestamp at the beginning of the call is when the callee picked up and typically the next timestamp is when audio begins playing, noted with a "Newly queued prompts are now being played" in the call log.

Regarding issue #3, to record the approximate time of when the VoiceXML script was fetched, you could place a <log> tag at the beginning of your script to log the time. There are timestamps throughout the call log that you can reference. Any log lines that run together occurred within the same second on the server.

Regarding issue #4, there were no system issues occurring between 11-12PM EST yesterday that would have caused this issue.

In order to fully involve our engineering team, it would be best for you to submit this issue as a ticket to support@plumgroup.com, along with any specific call logs that demonstrate this issue. Please remember to include your 8-digit customer ID in the subject line. This 8-digit customer ID can be found in your support contract with Plum.

Hope this helps.

Regards,
Plum Support