Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.
This project is now read‑only. Starting Monday, February 2, please use https://ibm-ceph.atlassian.net/ for all bug tracking management.

Bug 1595874

Summary: garbage collection takes a *long* time to get going
Product: [Red Hat Storage] Red Hat Ceph Storage Reporter: John Harrigan <jharriga>
Component: RGWAssignee: Matt Benjamin (redhat) <mbenjamin>
Status: CLOSED NOTABUG QA Contact: ceph-qe-bugs <ceph-qe-bugs>
Severity: medium Docs Contact: Aron Gunn <agunn>
Priority: unspecified    
Version: 3.1CC: agunn, cbodley, ceph-eng-bugs, dfuller, hnallurv, jbrier, jharriga, kbader, kdreyer, mbenjamin, owasserm, sweil, vakulkar
Target Milestone: rcFlags: vakulkar: automate_bug?
Target Release: 3.2   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Known Issue
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2018-10-02 20:11:38 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
fillWorkload.xml
none
emptyWorkload.xml
none
fillWorkload logfile
none
empty workload logfile with GC pending progress none

Description John Harrigan 2018-06-27 15:58:03 UTC
Created attachment 1455075 [details]
fillWorkload.xml

Description of problem:
My workload fills the cluster (using init and prepare operations) and then empties, using 'cleanup' and 'dispose' operations. The cluster is filled to
30 % RAW. My script monitors %RAW USED and pending GCs after the empty
workload is completed. The number of Pending GC's does not reduce for nearly
two hours after the empty workload completes, on an otherwise idle cluster!
See the two attached XML files.

Version-Release number of selected component (if applicable):
RHCEPH-3.1-RHEL-7-20180530.ci.0

How reproducible:
always

Steps to Reproduce:
1. fill cluster to 30%
2. empty cluster
3. no workload applied to cluster
4. monitor %RAW USED and pending GC's
5. Note that GC doesn't start for a long delayed period (1hr 40min)

Actual results:
GC doesn't start for a long delayed period (1hr 40min)

Expected results:
GC should start much sooner

Comment 3 John Harrigan 2018-06-27 15:59:41 UTC
Created attachment 1455076 [details]
emptyWorkload.xml

Comment 4 John Harrigan 2018-06-27 16:25:28 UTC
Created attachment 1455080 [details]
fillWorkload logfile

Comment 5 John Harrigan 2018-06-27 16:26:31 UTC
Created attachment 1455082 [details]
empty workload logfile with GC pending progress