Re: [IOT FND] [API - Issue] Delay between "issueLastUpdateTime" and query avaibility ?

ypanier · ‎12-30-2020

Hello,

I have a script who poll every minute the issue endpoint to forward the events to an external tool.
The only query parameter is 'lastUpdateTime>XXXX-XX-XX XX:XX:XX:XX:000' (where XXXX-XX-XX XX:XX:XX:XX:000 is the last time we poll the api).

And it seems that sometimes the script miss some events for an obscure reason, if I restart the script with a timestamp before the missed events, the events are seen and send correctly to the external tool.

Since the script poll the api every minute, i think that the events are not returned by the api directly but there is a small delay ( 30s ? 1min ?) between the "lastUpdateTime" and the availability by the api.

Example of what i think happen :

[2020-12-30 04:00:00] the event is closed at "2020-12-30 04:00:00"
[2020-12-30 04:00:30] a query with "lastUpdateTime>2020-12-30 03:59:30" doesn't return the event but should have.
[2020-12-30 04:01:30] the same query is send and return the closed event now (small delay between the closed issue and availability in api ?). The closed event has "issueLastUpdateTime=2020-12-30 04:00:00" so it should have been returned in the previous query.

Is there someone who could confirm me that there is a delay between the "lastUpdateTime" field value returned by the API and its availability through the API please ?
Does this delay is hardcoded ? (batch update) or does this delay depend on external factors like the load average of the iot fnd VM etc.. ?

Thanks,

Best Regards,

maiyub · ‎01-05-2021

Although in FND code, there isn't any delay configured for the above scenario, there will be some delays internally but shouldn't be a lot.

The factors affecting it are:
1) Is the script running on the same server as FND? If so, there could be a small delay between event time and committing to the database
2) if script is running on a separate server, is it time synced to the same source as FND server.
3) is the issue being closed manually (in GUI by user intervention) or automatically (by backend code). time to process and commit to DB may vary slightly.

ypanier · ‎03-25-2021

Sorry for the delay.

@maiyub wrote:
Although in FND code, there isn't any delay configured for the above scenario, there will be some delays internally but shouldn't be a lot.

Even with a huge load when there is many equipment and many events ?
Seems from my understanding after reading what you wrote that there is no batch mechanism ?
I was thinking you will do batch write since it could be receive many events/data and avoid bottleneck on writes to the database.

The factors affecting it are:
1) Is the script running on the same server as FND? If so, there could be a small delay between event time and committing to the database

The script is running on another server but there is low latency between them.
And even if there is a latency, the timestamp is initialize at the very beginning of the script and should not be part of the problem I think.

2) if script is running on a separate server, is it time synced to the same source as FND server.

Both are synced with the same NTP server

3) is the issue being closed manually (in GUI by user intervention) or automatically (by backend code). time to process and commit to DB may vary slightly.

No, the affected issues was automatically closed.