Bug 1734700

Summary:	rbd map is taking more than 2 minutes to map the device on the node when there are more parallel requests
Product:	[Red Hat Storage] Red Hat Ceph Storage	Reporter:	Madhu Rajanna <mrajanna>
Component:	RBD	Assignee:	Ilya Dryomov <idryomov>
Status:	CLOSED NOTABUG	QA Contact:	Madhavi Kasturi <mkasturi>
Severity:	high	Docs Contact:
Priority:	high
Version:	4.0	CC:	assingh, ceph-eng-bugs, hchiramm, jdillama
Target Milestone:	rc
Target Release:	4.0
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2019-09-04 17:52:28 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 Madhu Rajanna 2019-07-31 09:00:51 UTC

Description of problem:

rbd map is taking more than 2 minutes to map the device on the node when there are 50 parallel requests

when have seen this behavior in ceph-csi


Version-Release number of selected component (if applicable): ceph 12.2.2


How reproducible:
Always

Steps to Reproduce:
1.create 50/100 rbd images
2. map 50 rbd images on the node (parallel request)

Actual results:

for some requests, rbd map is taking more than 2 minutes to map images on node

Expected results:

rbd map should be completed in few seconds even if there are N number for parallel requests

Additional info:

link to the upstream issue https://github.com/ceph/ceph-csi/issues/507
there is a discussion is going on in upstream, creating this bug to track the same.

Comment 1 Jason Dillaman 2019-09-04 17:52:28 UTC

This was due to an extremely resource constrained environment and extra IO induced by running mkfs.ext4. The RBD CSI driver has switched to using XFS as its default and the problem seems to be better as per the linked issue.