[volume] Graceful restart volumes during active recording

[kmlebedev]

Describe the bug

After restarting all volumes sequentially on dc1 I catch errors

E0713 07:33:27     1 filer_server_handlers_read.go:168] failed to stream content /buckets/xxxxx-store/files/2021/07/13/e4a7348c-19de-489b-b78a-9cce9ea4a890: read chunk: http://fast-volume-2.s3-fast-volume.dc1:8080/110,0573bbc022d905?readDeleted=true: 404 Not Found
E0713 07:33:27     1 filer_server_handlers_read.go:168] failed to stream content /buckets/xxxxx-store/files/2021/07/13/e4a7348c-19de-489b-b78a-9cce9ea4a890: read chunk: http://fast-volume-2.s3-fast-volume.dc1:8080/110,0573bbc022d905?readDeleted=true: 404 Not Found

then a catch catch on dc2

E0713 09:27:42     1 filer_server_handlers_read.go:168] failed to stream content /buckets/xxxxx-store/files/2021/07/13/117b4ac7-c197-430b-abf4-c6046a51a0ec: read chunk: http://fast-volume-0.s3-fast-volume.dc2:8080/112,0593aec3661828?readDeleted=true: 404 Not Found
E0713 09:27:53     1 filer_server_handlers_read.go:168] failed to stream content /buckets/xxxxx-store/files/2021/07/13/117b4ac7-c197-430b-abf4-c6046a51a0ec: read chunk: http://fast-volume-0.s3-fast-volume.dc2:8080/112,0593aec3661828?readDeleted=true: 404 Not Found

System Setup

2.53
/usr/bin/weed volume -preStopSeconds=15

Expected behavior
Graceful restart volumes during active recording without errors

[chrislusf]

Added a fix to always fsync when graceful shutting down.