Bug 935553

Summary:	problems in mdcheck
Product:	[openSUSE] openSUSE Distribution	Reporter:	Ulrich Windl <Ulrich.Windl>
Component:	Basesystem	Assignee:	Coly Li <colyli>
Status:	RESOLVED FIXED	QA Contact:	E-mail List <qa-bugs>
Severity:	Enhancement
Priority:	P5 - None	CC:	mpluskal, nfbrown
Version:	13.2
Target Milestone:	---
Hardware:	All
OS:	openSUSE 13.2
Whiteboard:
Found By:	---	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---

Description Ulrich Windl 2015-06-22 07:02:46 UTC

While working with openSUSE 13.2 late at night I realized that the system suddenly became very slow. Some moment later I realized why: mdcheck had started a RAID check on my RAID1, causing heavy I/O and massive read delays.

While examining the problem I found out these defects of mdcheck:
1) mdcheck does not write to syslog when a) it starts checking a RAID, b) when it continues checking a RAID, c) when it finished checking a RAID, d) when it was interrupted while cheking a RAID

2) mdcheck does not allow to limit the I/O bandwidth while performing the check

3) mdcheck does not cleanup when terminated by a signal (e.g. when the system is shut down). Specifically "tarp 0" does not do that; it's just for c "clean" exits. On effect of theis is that there are remaining temporary files left around.

Comment 1 Neil Brown 2017-10-06 01:37:00 UTC

Thanks for the suggestions.
I've submitted 1 and 3 upstream, so they should flow into suse in due course.
I'm not convinced 2 should be done in mdcheck, but I'm not really sure where that sort of setting should be imposed.  Ideally the defaults would always be suitable, but apparently not in your case.

Comment 2 Neil Brown 2018-01-04 05:38:58 UTC

hi Coly,
could you take this one too?  The following two commits should go to Factory when convenient. Maybe to 12-SP3 too??

Commit: f258a5e3911c ("mdcheck: improve cleanup")
Commit: ea835492b6ff ("mdcheck: add some logging.")

Thanks.

Comment 3 Coly Li 2018-01-15 12:06:57 UTC

(In reply to Neil Brown from comment #2)
> hi Coly,
> could you take this one too?  The following two commits should go to Factory
> when convenient. Maybe to 12-SP3 too??
> 
> Commit: f258a5e3911c ("mdcheck: improve cleanup")
> Commit: ea835492b6ff ("mdcheck: add some logging.")
> 
> Thanks.

Hi Neil,

Thanks for the hint.
These two patches are in Factory already, and I send maintenance request to SLE12-SP3:Update with ID 151164, waiting for response.

Coly Li

Comment 7 Coly Li 2018-02-06 07:25:02 UTC

Patches are accepted and picked as SUSE:Maintenance:5402.

Here I close this report.

Coly Li

Comment 8 Swamp Workflow Management 2018-06-05 10:14:47 UTC

SUSE-RU-2018:1500-1: An update that has 5 recommended fixes can now be installed.

Category: recommended (moderate)
Bug References: 1032802,1047314,1049126,1059596,935553
CVE References: 
Sources used:
SUSE Linux Enterprise Server 12-SP3 (src):    mdadm-4.0-6.11.1
SUSE Linux Enterprise Desktop 12-SP3 (src):    mdadm-4.0-6.11.1

Comment 9 Swamp Workflow Management 2018-06-07 22:09:51 UTC

openSUSE-RU-2018:1589-1: An update that has 5 recommended fixes can now be installed.

Category: recommended (moderate)
Bug References: 1032802,1047314,1049126,1059596,935553
CVE References: 
Sources used:
openSUSE Leap 42.3 (src):    mdadm-4.0-5.3.1