Bug 1378975

Summary: Swap observed due to puma on UI appliances, causing unresponsive UI
Product: Red Hat CloudForms Management Engine Reporter: Jerome Marc <jmarc>
Component: PerformanceAssignee: Nick LaMuro <nlamuro>
Status: CLOSED WONTFIX QA Contact: Pradeep Kumar Surisetty <psuriset>
Severity: medium Docs Contact:
Priority: high    
Version: 5.6.0CC: arcsharm, cpelland, dajohnso, dmetzger, gjansen, hroy, jhardy, jmarc, kmorey, nlamuro, nstephan, obarenbo
Target Milestone: GAFlags: nlamuro: needinfo? (dajohnso)
Target Release: cfme-future   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard: perf:ui
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2018-01-08 13:30:08 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: CFME Core Target Upstream Version:
Attachments:
Description Flags
Troubleshooting results for free, smem, smem -W and vmstat commands none

Description Jerome Marc 2016-09-23 18:03:49 UTC
Created attachment 1204244 [details]
Troubleshooting results for free, smem, smem -W and vmstat commands

Description of problem:
UI appliances become very slow after some time on the MBU Lab. Looking at the UI appliances, we observe memory swap happening, specifically due to puma. Restarting the appliance is the only fix.

Version-Release number of selected component (if applicable):
5.6.1.2.20160810181333_8ba817b 

How reproducible:
Always

Steps to Reproduce:
1. Have CloudForms setup in a Production environment (e.g. 2 UI appliances, 1 BD master, 9 worker appliances) 
2. Give access to users and let them use CloudForms

Actual results:
Over time, swap will start occurring on the UI appliance. Analysis seems to point at Puma (see command results attached).

Expected results:
Swap should not happen. These appliances are 8GB RAM, with 4 UI workers each.

Additional info:
Happy to share the environment credentials for further troubleshooting.

Comment 2 Jerome Marc 2017-01-23 22:24:00 UTC
Same issue experienced in 5.7.

Comment 17 dmetzger 2018-01-08 13:30:08 UTC
Closing this 5.6 ticket based on the work / investigation completed to the point. If the problem is found reproducible in a more current release a new ticket should be created. My understanding at this time is there have been problems reproducing this reliably on 5.6 and it has not be reported on newer releases.