Note: This is a beta release of Red Hat Bugzilla 5.0. The data contained within is a snapshot of the live data so any changes you make will not be reflected in the production Bugzilla. Also email is disabled so feel free to test any aspect of the site that you want. File any problems you find or give feedback here.
Bug 1514504 - [RFE] Implement rocksdb cache perf counter
Summary: [RFE] Implement rocksdb cache perf counter
Keywords:
Status: CLOSED WONTFIX
Alias: None
Product: Red Hat Ceph Storage
Classification: Red Hat
Component: RADOS
Version: 2.4
Hardware: x86_64
OS: All
high
high
Target Milestone: rc
: 3.*
Assignee: Josh Durgin
QA Contact: ceph-qe-bugs
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2017-11-17 15:35 UTC by Vikhyat Umrao
Modified: 2019-03-08 04:10 UTC (History)
4 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2019-03-08 04:10:14 UTC


Attachments (Terms of Use)


Links
System ID Priority Status Summary Last Updated
Ceph Project Bug Tracker 22152 None None None 2017-11-17 15:43:48 UTC

Description Vikhyat Umrao 2017-11-17 15:35:01 UTC
Description of problem:
Implement rocksdb cache perf counter

Maybe we can use ceph daemon osd.$id perf dump and add rocksdb cache perf counter:

1. rocksdb cache full or not
2. how many omap keys it has 
3. How many times it became full and flush?
4. any other important information

Version-Release number of selected component (if applicable):
Red Hat Ceph Storage 2.4


Right now perf dump covers the following:

  "rocksdb": {
        "rocksdb_get": 9798,
        "rocksdb_transaction": 518732,
        "rocksdb_get_latency": {
            "avgcount": 9798,
            "sum": 54.697089911
        },
        "rocksdb_submit_latency": {
            "avgcount": 505433,
            "sum": 15.398420482
        },
        "rocksdb_submit_sync_latency": {
            "avgcount": 13299,
            "sum": 12.566100383
        },
        "rocksdb_compact": 0,
        "rocksdb_compact_range": 0,
        "rocksdb_compact_queue_merge": 0,
        "rocksdb_compact_queue_len": 0
    },

Maybe we can add rocksdb cache here.

Comment 2 Greg Farnum 2019-03-08 04:10:14 UTC
We haven't ever wanted this information, including when Mark built the BlueStore auto-tuning cache stuff. The specifically-named values don't make much sense either — it is in the nature of an SST cache to always be full and flushing something (especially under compaction); it doesn't track specific key counts but rather ranges of level files; etc.


Note You need to log in before you can comment on or make changes to this bug.