Bug 199252 - vmstat -d gives inaccurate results on idle disks.
vmstat -d gives inaccurate results on idle disks.
Status: CLOSED INSUFFICIENT_DATA
Product: Red Hat Enterprise Linux 4
Classification: Red Hat
Component: procps (Show other bugs)
4.0
x86_64 Linux
medium Severity medium
: ---
: ---
Assigned To: Tomas Smetana
Brian Brock
:
Depends On:
Blocks:
  Show dependency treegraph
 
Reported: 2006-07-18 09:25 EDT by Bret Hammond
Modified: 2008-08-02 19:40 EDT (History)
0 users

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Environment:
Last Closed: 2007-08-02 03:48:05 EDT
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)
output from vmstat -d and sar -d over the same time interval. (3.57 KB, application/x-gzip)
2006-07-18 09:25 EDT, Bret Hammond
no flags Details

  None (edit)
Description Bret Hammond 2006-07-18 09:25:38 EDT
Description of problem:

vmstat -d <delay> <count> shows activity on disks when there is none.

These disk are attached to  Data Direct S2A9500 controllers.  The
quiesced state of the disks has been verified via ddns stats command
on the controllers, and via a sar -d run during the same interval.


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

This is RHEL 4 WS x86_64 using procps-3.2.3-8.3


How reproducible:

This has been reproduced on every node that I have tried it on.  Just
run vmstat -d 5 5, while verifying no disk activity via the ddn 
controllers, or sar -d 5 5.  Relevant output attached.  /dev/sda
is the only sd device that should show activity.


Steps to Reproduce:
1.
2.
3.
  
Actual results:


Expected results:


Additional info:
Comment 1 Bret Hammond 2006-07-18 09:25:38 EDT
Created attachment 132601 [details]
output from vmstat -d and sar -d over the same time interval.
Comment 2 Karel Zak 2006-08-07 07:50:15 EDT
The source data for vmstat is /proc/diskstats. You can try check this file if
you think that vmstat outputs unreliable information. 

The /proc/diskstats is I/O statistics (history) -- so these numbers mean that
there was an activity, but it says nothing about actual activity. You need to
check (compare) statistics from more loops.

I don't see any progress for sdb I/O activity in your output, I see changes for
sda only:

sda    28188   7509 1006241  123646  80442  99410 1438784 1917445      0    131
sdb     4841   1669   35064   19417 219011 19747547 50370193 1356765620      0 
  267

sda    28188   7509 1006241  123646  80447  99429 1438976 1917447      0    131
sdb     4841   1669   35064   19417 219011 19747547 50370193 1356765620      0 
  267

sda    28188   7509 1006241  123646  80451  99435 1439056 1917448      0    131
sdb     4841   1669   35064   19417 219011 19747547 50370193 1356765620      0 
  267

Is it more clear now?


Comment 3 Tomas Smetana 2007-08-02 03:48:05 EDT
Nobody responded to this for almost a year... Closing.

Note You need to log in before you can comment on or make changes to this bug.