[SERVER-25180] Make latency histogram collection optional in full-time diagnostic data capture Created: 20/Jul/16 Updated: 21/May/18 Resolved: 29/Aug/16 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Diagnostics, Querying |
| Affects Version/s: | None |
| Fix Version/s: | 3.3.12 |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Bruce Lucas (Inactive) | Assignee: | Kyle Suarez |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||||||||||
| Sprint: | Integration 2016-09-19 | ||||||||||||||||
| Participants: | |||||||||||||||||
| Description |
|
The new latency histogram facility adds a considerable amount of data to ftdc, decreasing the retention period significantly. It has limited value for the primary purpose of always-on ftdc, because mongod logs provide information about the high-latency operations which are generally of interest for after-the-fact issue analysis. Collection of latency histogram data by ftdc should be optional, off by default. It should be possible to enable capture of latency histogram data in ftdc dynamically at runtime if desired to use it for targeted performance investigations. |
| Comments |
| Comment by Kyle Suarez [ 29/Aug/16 ] | ||
|
Marking Documentation Changes as "needed", since this changes the previous API for the histogram-related commands. | ||
| Comment by Githook User [ 29/Aug/16 ] | ||
|
Author: {u'username': u'kevinAlbs', u'name': u'Kevin Albertson', u'email': u'kevin.albertson@10gen.com'}Message: For full per-collection histograms, use the aggregation stage
For the full global histogram, invoke serverStatus with
Branch: master | ||
| Comment by Kyle Suarez [ 19/Aug/16 ] | ||
|
Putting this back on the backlog since kevin.albertson's internship is coming to a close. Because of the downstream impact of this ticket, I'll try to get this done as soon as possible in the next iteration. | ||
| Comment by Bruce Lucas (Inactive) [ 01/Aug/16 ] | ||
|
Note by the way that the overall stats "ops" and "latency" for each category should be retained as this is a small amount of data and will allow computing average latency which is very useful information. |