Bug 1704705

Summary: [RHGS] glusterfs high memory use - memory availability question.
Product: [Red Hat Storage] Red Hat Gluster Storage Reporter: Andrew Robinson <anrobins>
Component: fuseAssignee: Amar Tumballi <atumball>
Status: CLOSED WORKSFORME QA Contact: Rahul Hinduja <rhinduja>
Severity: low Docs Contact:
Priority: unspecified    
Version: rhgs-3.4CC: amukherj, nchilaka, rhs-bugs, sankarshan, storage-qa-internal
Target Milestone: ---Keywords: Question
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2019-05-21 05:08:14 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:

Description Andrew Robinson 2019-04-30 11:58:11 UTC 
Description of problem:

This is a general question. I have several support cases recently where a glusterfs process on a client used a very high percentage of the system memory. My understanding is these instances fall into two groups, memory leaks, and events that cause legitimate high memory usage. 

My question is about the second group, events, such as blocked IO or an application that sucks up resources, that cause high memory usage. When these events complete or terminate, the glusterfs process still retains the memory. 

Is that memory locked away by the glusterfs process, or can the kernel reap the memory and allocate it to other processes? 


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

RHGS 3.4 release 0


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

If I can tell a customer that the memory apparently assigned to the glusterfs process can be recovered and used by other processes, that would alleviate a lot of anxiety on the part of that customer.
Comment 3 Amar Tumballi 2019-05-03 08:05:00 UTC 
We had issues earlier with memory consumed by process not getting released. With 3.4.4 (Batch Update 4), we implemented a Garbage Collection logic in FUSE module, which now releases memory if not used from some time (lru-limit option). 

So, any customers facing memory consumption related errors/issues, please ask them to upgrade their clients to RHGS 3.4.4 (glusterfs-3.12.2_47+)
Comment 7 Amar Tumballi 2019-05-21 05:08:14 UTC 
Marking as WORKSFORME with information on comment#3.