Bug 1279490

Summary: [RFE] Post concurrent sync of full repo content, " hammer repository synchronize" command takes more than 50% time to reach 100%.
Product: Red Hat Satellite Reporter: Pradeep Kumar Surisetty <psuriset>
Component: RepositoriesAssignee: satellite6-bugs <satellite6-bugs>
Status: CLOSED NOTABUG QA Contact: Katello QA List <katello-qa-list>
Severity: high Docs Contact:
Priority: unspecified    
Version: 6.1.4CC: bbuckingham
Target Milestone: UnspecifiedKeywords: FutureFeature, Triaged
Target Release: Unused   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Enhancement
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2017-06-01 04:24:29 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
Network l2 in Mbits per sec
none
disk iops none

Description Pradeep Kumar Surisetty 2015-11-09 14:59:35 UTC
Created attachment 1091779 [details]
Network l2 in Mbits per sec

Description of problem:

1) Installed Satelitte 6.1.4 compose 5. 
2) Sync 4 repos concurrently  (total content size: 104GB) from local repo server. ( created a local repo with CDN content) 
 
       Red Hat Enterprise Linux 5 Server RPMs i386 5Server 
       Red Hat Enterprise Linux 6 Server RPMs x86_64 5Server
       Red Hat Enterprise Linux 5 Server RPMs i386 5Server
       Red Hat Enterprise Linux 6 Server RPMs x86_64 5Server

3) even though "hammer repository synchronize " command continue to run till it reaches 100%, all content is copied almost 20 mins before itself.   

Essentially "hammer repository synchronize " command spend more than 50% of time after syncing the complete data ( in my case 104G). 

Attached Network l2_Mbits per sec and DISK IOPS graphs. As shown in graphs, post content syncing there is no network transfer or disk IOPS. But command (hammer repository synchronize) continue to run more than 50% of total time untill it reaches 100%   


version-Release number of selected component (if applicable):

pulp-server-2.6.0.17-1.el7sat.noarch
katello-2.2.0.16-1.el7sat.noarch
candlepin-0.9.49.9-1.el7.noarch
rubygem-hammer_cli-0.1.4.13-1.el7sat.noarch
satelitte 6.1.4 

How reproducible:



Steps to Reproduce:
1. Concurrently sync repos from local repo server to satelitte server as shown above. 
2.
3.

Actual results:


Expected results:


Additional info:

Comment 1 Pradeep Kumar Surisetty 2015-11-09 15:00:27 UTC
Created attachment 1091780 [details]
disk iops

Comment 3 Brian Bouterse 2016-03-21 20:16:38 UTC
There are several ways that metadata processing could be sped up but its such a big problem there are probably many small things we could do to make improvements. Consider filing a story to speed up metadata handling on the upstream Pulp issue tracker [0].

Regarding hammer waiting at 100% for 20 minutes, could hammer not be reporting on the task status correctly. A sync usually triggers one or more publishes after it. Perhaps hammer is not following the subsequently generated tasks whose IDs are in the spawned_task field?

[0]: https://pulp.plan.io/