[SERVER-5828] Metric/Stats Tracking Created: 11/May/12  Updated: 05/May/17  Resolved: 05/May/17

Status: Closed
Project: Core Server
Component/s: Admin
Affects Version/s: None
Fix Version/s: None

Type: Improvement Priority: Major - P3
Reporter: Scott Hernandez (Inactive) Assignee: Unassigned
Resolution: Done Votes: 11
Labels: metrics, mongostat, stats, top
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
Related
related to TOOLS-42 Make mongostat configurable Closed
related to SERVER-44432 Per index access stats Closed
related to SERVER-13220 Add and expose statistics to new quer... Backlog
related to SERVER-2178 Track stats per db/collection Closed
related to SERVER-3109 Expose index stats Closed
related to SERVER-4202 Add a way to get stats on number of q... Closed
related to SERVER-4897 make JournalStats in ServerStatus cou... Closed
related to SERVER-5787 Track opcounters (insert/delete/updat... Closed
related to SERVER-5173 Track replication getMores separately... Closed
related to SERVER-5784 Record all btree stats (stop sampling) Closed
is related to SERVER-5905 Add data collection and command to ge... Closed
is related to SERVER-1740 Add rate limiting to creation of syst... Backlog
is related to SERVER-1739 trace total nscanned and other query/... Closed
is related to SERVER-4672 add monitoring of version reload stat... Closed
is related to SERVER-7081 Track (new) replication stats Closed
is related to SERVER-7382 Track counters for each command Closed
is related to SERVER-12172 Add finer granularity metrics for ser... Closed
Backwards Compatibility: Fully Compatible
Participants:

 Description   

Issues to bring together metric and stats related changes for making the server better instrumented and increase visibility into internal rates/metrics.



 Comments   
Comment by Ian Whalen (Inactive) [ 05/May/17 ]

We no longer use SERVER tracking tickets in this way. Specific improvement requests should be requested individually.

Comment by Grégoire Seux [ 20/Dec/12 ]

This would be really great to gain more dba support toward mongodb. Right now tooling and metrics are at a bare minimum.

Comment by Daniel Pasette (Inactive) [ 26/Oct/12 ]

Author:

{u'date': u'2012-10-26T15:12:12-07:00', u'name': u'Eliot Horowitz', u'email': u'erh'}

Message: SERVER-5828 - rewrite sererStatus caommands to be modular
allow any code to register sections or metrics without touching core code
have same core serverStatus code for mongos and mongod
Branch: master
https://github.com/mongodb/mongo/commit/b56566bf0ec3cb00221e8e951b24f47783ec2af3

Comment by Scott Hernandez (Inactive) [ 31/Aug/12 ]

Track min/max/avg ms per query (template) or index; maybe easier to track this at the collection level for all queries on that collection to start with.

Comment by Scott Hernandez (Inactive) [ 13/Aug/12 ]

Add counters to track documents touched/returned grouped by opcounter, and total.

For example, this would allow us to answer the number of documents affected by queries/updates/removes/commands/etc.

Comment by Scott Hernandez (Inactive) [ 07/Aug/12 ]

Add counters about query planner (multiplan execution) and how often it is run on each collection/index.

Comment by Scott Hernandez (Inactive) [ 11/May/12 ]

Other notes/ideas from 5/11/12 meeting:

Track counters by source (mongos or direct, user initiated, system – repl/shard-cmd --, localhost/ip, user)

Ind. cmd (types) by coll/db – like count, aggregate, mapReduce, drop/etc

Histograms/buckets by costs/latency and variance for things

Need nano/microsecs timing with reporting in fractions of ms/micros

Track extent stuff, freelist/etc., create/delete, per file

Data size w/without padding.

Maybe track possible faults per ns.

Track network related stats: create/destroy, socket exceptions, outgoing bytes, buffer bytes, sockets stats (ops, bytes, stack usage)

BSON* operations: iteration, append, realloc, copy

Update operations: ex $addToSet/$pull/$etc.

How to answer if fragment exists and compact is needed.

Comment by Scott Hernandez (Inactive) [ 11/May/12 ]

Other ideas:

Most frequent/slowest by period:

  • Slowest ops per coll/db by type (top N/Mmin)
  • Highest occurrence by query template (top N/Mmin)
  • Most used indexes (top N/Mmin)

...

Generated at Thu Feb 08 03:09:59 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.