Repository navigation
1.8.2
OK, so I was watching the Staging server and notice that a series of network operations (about 8) were continually responding with ECONNRESET or PROTOCOL_PACKETS_OUT_OF_ORDER.
The result was our retry() function would immediately try again.
As things were continuing, the CPU % was > 100 ( 106 - 124 ) continually. It's possible that having the % CPU so high might have contributed to the problem.
This patch will try to introduce some delays and a limit to the retry() process.
The delays will prevent the % CPU from spiking as much and thus limiting any errors due to high CPU usage.
Also a limit will prevent this from being a constant drain on the process. Most of these network problems should be one off or two off kinds of events. If we are trying for 10 times or more, then this isn't a simple network connection issue.
Release Notes
- [wip] introduce a delay and a limit to our retry() operations