If the memory copy is not completed, the worker process reload will lose the data.

[barnettZQG]

In the automated process, changes to dyups or nginx reload may occur continuously. In such cases, we find that the data modified by dyups is invalid.
exapmle the following log:

2018/10/15 15:09:59 [notice] 1#0: signal 29 (SIGIO) received
2018/10/15 15:09:59 [notice] 1#0: signal 17 (SIGCHLD) received from 932
2018/10/15 15:09:59 [notice] 1#0: worker process 932 exited with code 0
2018/10/15 15:09:59 [notice] 1#0: signal 29 (SIGIO) received
2018/10/15 15:09:59 [notice] 1#0: signal 17 (SIGCHLD) received from 936
2018/10/15 15:09:59 [notice] 1#0: worker process 936 exited with code 0
2018/10/15 15:09:59 [info] 934#0: [dyups] create upstream 5000.grf5b07e.yaufe6r5.tvga8.goodrain.org
2018/10/15 15:09:59 [notice] 1#0: signal 29 (SIGIO) received
2018/10/15 15:09:59 [info] 934#0: [dyups] sync add: 5000.grf5b07e.yaufe6r5.tvga8.goodrain.org rv: success rc: 200
2018/10/15 15:09:59 [info] 934#0: [dyups] delete upstream "8080.grf5b07e.yaufe6r5.tvga8.goodrain.org"
2018/10/15 15:09:59 [info] 934#0: [dyups] free dynamic upstream in find upstream 24
2018/10/15 15:09:59 [info] 934#0: [dyups] sync add: 8080.grf5b07e.yaufe6r5.tvga8.goodrain.org rv: success rc: 200
2018/10/15 15:09:59 [info] 934#0: [dyups] create upstream 3000.grf5b07e.yaufe6r5.tvga8.goodrain.org
2018/10/15 15:09:59 [info] 934#0: [dyups] sync add: 3000.grf5b07e.yaufe6r5.tvga8.goodrain.org rv: success rc: 200
2018/10/15 15:09:59 [notice] 934#0: exiting
2018/10/15 15:09:59 [notice] 934#0: exit
2018/10/15 15:09:59 [notice] 1#0: signal 17 (SIGCHLD) received from 934
2018/10/15 15:09:59 [notice] 1#0: worker process 934 exited with code 0

[barnettZQG]

934 worker process status is waiting gracefully shutting down. but It also accepted the dyups request.