0038007: LogClient can generate too many useless requests if server is struggling

Openbravo Issue Tracking System - Retail Modules

View Issue Details

Project

Category

View Status

Date Submitted

Last Update

0038007

Retail Modules

Web POS

public

2018-02-27 11:39

2018-03-01 10:14

Reporter

marvintm

Assigned To

marvintm

Priority

urgent

Severity

major

Reproducibility

always

Status

closed

Resolution

fixed

Platform

OS Version

Product Version

Target Version

Fixed in Version

Merge Request Status

Review Assigned To

migueldejuana

OBNetwork customer

Support ticket

Regression level

Regression date

Regression introduced in release

Regression introduced by commit

Triggers an Emergency Pack

Summary

0038007: LogClient can generate too many useless requests if server is struggling

Description

Currently, the logclient is sent every 30 seconds, and the request has a timeout of 20 seconds. Normally, this is fine, and there is more than enough time to send even the maximum page size in these 20 seconds.

However, sometimes the backend server may be struggling for some other reason, or the bandwidth in the store may not be enough, and finally the request cannot be completed in this time. In this case, as the interval is currently 30 seconds, the terminal will every 30 seconds send another request, which most likely will also fail, thus eventually flooding the server with these requests, and wasting a significant amount of network bandwidth and server resources.

We don't want to change the interval because in normal cases we want to send the logclient information as soon as possible. However, we can increase the timeout of the logclient by default, and the terminal should behave smartly and should not generate another request if the previous one is still being processed.

With these two changes, the terminal will no longer flood the server with requests in case it's struggling, but instead will automatically adjust itself, and stop sending the logclient information if the server is not able to absorb it.

Steps To Reproduce

- Create a problem in the server when accepting the logclient (for example, add a breakpoint in LogClientLoader class, or artificially reduce the available bandwidth very significantly).
- Wait for the logclient request. Verify that it fails due to timeout.
- Verify that 10 seconds, later, another request is done, which will also fail in 20 seconds, and 10 seconds later another one will be done, and 20 seconds later it will fail, ...

This was the intended behaviour, but we have now seen that this can create significant problems in an already struggling server,

Proposed Solution

- Timeout in logclient request will be significantly increased.
- LogClient process in the terminal will be changed, so that if there is a pending request, a new request will not be triggered, and instead the process will wait for the next cycle to check if a new request can now be done.

Additional Information