[SERVER-48131] Report completion and duration of downgrading WT datafiles step Created: 12/May/20  Updated: 29/Oct/23  Resolved: 20/May/20

Status: Closed
Project: Core Server
Component/s: Diagnostics, Storage
Affects Version/s: None
Fix Version/s: 4.4.0-rc7, 4.7.0

Type: Task Priority: Major - P3
Reporter: Bruce Lucas (Inactive) Assignee: Daniel Gottlieb (Inactive)
Resolution: Fixed Votes: 1
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Backports
Backwards Compatibility: Fully Compatible
Backport Requested:
v4.4
Sprint: Execution Team 2020-06-01
Participants:

 Description   

The following reports the beginning of an operation that can be very lengthy. It would be helpful for diagnosis to report the completion with a "durationMillis" attr.

{"t":{"$date":"2020-05-12T07:22:57.433Z"},"s":"I","c":"STORAGE","id":22324,"ctx":"conn108695","msg":"Downgrading WiredTiger datafiles.","source":{"sn":1,"ln":37327}}



 Comments   
Comment by Githook User [ 20/May/20 ]

Author:

{'name': 'Daniel Gottlieb', 'email': 'daniel.gottlieb@mongodb.com', 'username': 'dgottlieb'}

Message: SERVER-48131: Additional logging and timing for interesting WT startup and shutdown events.

(cherry picked from commit eae377f7f236fdb02889a96c983035d6f51dac6e)
Branch: v4.4
https://github.com/mongodb/mongo/commit/c97dc0af795864a77f1096a6a0aa8798142723ce

Comment by Githook User [ 20/May/20 ]

Author:

{'name': 'Daniel Gottlieb', 'email': 'daniel.gottlieb@mongodb.com', 'username': 'dgottlieb'}

Message: SERVER-48131: Additional logging and timing for interesting WT startup and shutdown events.
Branch: master
https://github.com/mongodb/mongo/commit/eae377f7f236fdb02889a96c983035d6f51dac6e

Comment by Bruce Lucas (Inactive) [ 12/May/20 ]

That makes sense. It seems that wt open can also be slow so it would be useful to add duration logging for that.

Also in each case we should log both the start and end (with duration) because when something like that is really slow it may not finish to log the end, e.g. due to oom or kill -9, but if we also log the start we will know what was in progress when the log ends abruptly.

Comment by Eric Milkie [ 12/May/20 ]

It's likely that the checkpoint on close is what is taking a long time. We should add duration logging for all calls to connection->close(), including this one for Downgrading.

Generated at Thu Feb 08 05:16:14 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.