-
Type:
Improvement
-
Resolution: Done
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: Logging
-
Labels:None
-
Storage Engines
-
(copied to CRM)
It has been observed in the most recent versions of MongoDB (4.4.5) that startup recovery of an instance with a large data set can take several hours.
This is problematic for TSEs and for the users as the status of the mongod process is not entirely clear.
For example:
- There is no indication how many phases or steps the recovery process has in total
- Where the current progress is
- How many steps are outstanding
To illustrate, it's been observed that a mongod spends multiple hours recovering after having a message similar to the following printed in the log with no other clear signs of progress:
{"t":{"$date":"2021-04-21T00:51:14.703+00:00"},"s":"I", "c":"STORAGE", "id":22430, "ctx":"initandlisten","msg":"WiredTiger message","attr":{"message":"[1618966274:703153][5978:0x7f67cef96bc0], file:collection-2-1893086824266225355.wt, txn-recover: [WT_VERB_RECOVERY_PROGRESS] Recovering log 62628 through 62628"}}
{"t":{"$date":"2021-04-21T00:51:14.759+00:00"},"s":"I", "c":"STORAGE", "id":22430, "ctx":"initandlisten","msg":"WiredTiger message","attr":{"message":"[1618966274:759611][5978:0x7f67cef96bc0], file:collection-2-1893086824266225355.wt, txn-recover: [WT_VERB_RECOVERY | WT_VERB_RECOVERY_PROGRESS] Set global recovery timestamp: (1618965463, 1)"}}
{"t":{"$date":"2021-04-21T00:51:14.759+00:00"},"s":"I", "c":"STORAGE", "id":22430, "ctx":"initandlisten","msg":"WiredTiger message","attr":{"message":"[1618966274:759674][5978:0x7f67cef96bc0], file:collection-2-1893086824266225355.wt, txn-recover: [WT_VERB_RECOVERY | WT_VERB_RECOVERY_PROGRESS] Set global oldest timestamp: (1618965458, 1)"}}
