-
Type: Task
-
Resolution: Unresolved
-
Priority: Major - P3
-
None
-
Affects Version/s: 3.6.3
-
Component/s: mongorestore
-
Labels:None
-
3
-
1,676
We've gotten an internal report of a restore continuing, but continuing to have the progress bar count up even though there are no more inserts.
Possibly steps to replicate:
- mongorestore several large collections with a direct connection to a replica set primary
- fail the primary during a long-running restore
Look for output like
2018-04-07T03:59:51.063-0500 [########................] dbName.collName 469GB/1278GB (36.7%) 2018-04-07T03:59:54.064-0500 [########................] dbName.collName 470GB/1278GB (36.8%) 2018-04-07T03:59:56.778-0500 error: EOF 2018-04-07T03:59:57.063-0500 [########................] dbName.collName 470GB/1278GB (36.8%) 2018-04-07T03:59:58.840-0500 error: EOF 2018-04-07T03:59:58.872-0500 error: EOF 2018-04-07T03:59:58.887-0500 error: EOF 2018-04-07T03:59:58.973-0500 error: EOF 2018-04-07T04:00:00.064-0500 [########................] dbName.collName 470GB/1278GB (36.8%)
I'm suspicious of the code at this point; I think it should return after sending the error to terminate iteration. However, that alone shouldn't be the culprit since {[db.IsConnectionError(err)}} should be true and we should never reach the log statements.