Bug 1651098

Summary:	dht_readdirp_cbk : Set inode for dir entries if inode is already linked.
Product:	[Community] GlusterFS	Reporter:	Nithya Balachandran <nbalacha>
Component:	distribute	Assignee:	Nithya Balachandran <nbalacha>
Status:	CLOSED NOTABUG	QA Contact:
Severity:	high	Docs Contact:
Priority:	high
Version:	mainline	CC:	bugs
Target Milestone:	---
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:	2019-11-04 04:24:44 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 Nithya Balachandran 2018-11-19 08:21:35 UTC

Description of problem


dht_readdirp_cbk sets the entry->inode to NULL for directory entries in order to force a lookup. This causes a massive slowdown when listing directories which contain a lot of sub-directories.

If the inode for the directory is already present in the inode table, it implies that the inode was linked (ie, a lookup was performed already). If the inode for the directory entry is already linked, we do not need to send another lookup for the directory.




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


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

Comment 1 Worker Ant 2018-11-19 08:46:41 UTC

REVIEW: https://review.gluster.org/21673 (cluster/dht: Set the entry->inode if dir was lookup up earlier) posted (#1) for review on master by N Balachandran

Comment 2 Amar Tumballi 2019-06-20 04:50:04 UTC

the above patch got abandoned due to inactivity timeout, and hence moving the bug to NEW.

Comment 3 Nithya Balachandran 2019-11-04 04:24:44 UTC

This can cause stale data to be saved in the cache. Closing the BZ.