Details
-
Task
-
Resolution: Fixed
-
Major
-
None
-
Security Level: Public
-
None
-
CBG Sprint 41
-
2
Description
The logging around 2.x replication code emits EOF errors at error level, which are expected to happen when a client disconnects from Sync Gateway.
2020-02-03T06:00:22.236Z [ERR] c:[15e93ddd] Invalid response to 'changes' message: RPY#3 -- EOF. Body: -- rest.(*blipSyncContext).Logf() at blip_sync.go:222
|
These can happen quite often, and there's nothing actionable about them. We should drop the log level specifically for EOF errors down to debug, given we already have info-level logging for client disconnects. We need to retain existing error level logging for all other errors.
Full log excerpt:
2020-02-03T06:00:22.187Z [INF] Cache: c:[15e93ddd] GetCachedChanges("user:5949a4d051648462805b7382f5595b659a42fa3f77432c4296b31e91293216c4", 8407579) --> 1 changes valid from #8407480
|
2020-02-03T06:00:22.191Z [INF] Cache: Received #8407583 (unused sequence)
|
2020-02-03T06:00:22.191Z [INF] Cache: Deferring #8407583 (1 now waiting for #8407582...#8407582) doc "" / ""
|
2020-02-03T06:00:22.201Z [INF] Cache: Received #8407582 (unused sequence)
|
2020-02-03T06:00:22.235Z [INF] Sync: c:[15e93ddd] Sent 1 changes to client, from seq 8407581
|
2020-02-03T06:00:22.236Z [INF] HTTP: c:[15e93ddd] #19299: --> BLIP+WebSocket connection closed
|
2020-02-03T06:00:22.236Z [ERR] c:[15e93ddd] Invalid response to 'changes' message: RPY#3 -- EOF. Body: -- rest.(*blipSyncContext).Logf() at blip_sync.go:222
|
2020-02-03T06:00:22.236Z [INF] Changes: c:[15e93ddd] MultiChangesFeed done (to 5949a4d051648462805b7382f5595b659a42fa3f77432c4296b31e91293216c4)
|