0

The I/O usage is high on drives of drdb, but the usage is very low on the underlying drive. Is it right to ignore usage of I/O on DRDB drive as the underlying disk usage is fine.

We have 4 Redhat Boxes Dell PowerEdge R630 (say a,b,c,d) having the following OS/packages.

RedHat EL 6.5 MySql Enterprise 5.6 DRBD 8.4 Corosync 1.4.7

We have setup 4-way stacked drbd resources as below:

Cluster Cluster-1: servers a and b connected to each other local lan Cluster Cluster-2: servers c and d

Cluster Cluster-1 and Cluster-2 are attached via stacked drbd via virtual IPs and are part of different data centres.

The drbd0 disks have been created locally on each servers 1GB, and are further attached to drbd10.

Underlying Disk usage that is normal:-

    Device:         rrqm/s   wrqm/s     r/s     w/s   rsec/s   wsec/s avgrq-sz avgqu-sz   await  svctm  %util
sda2              0.00  3718.00    0.00  441.00     0.00 33268.00    75.44     0.06    0.13   0.08   3.35

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           1.25    0.00    0.81    1.75    0.00   96.18

Device:         rrqm/s   wrqm/s     r/s     w/s   rsec/s   wsec/s avgrq-sz avgqu-sz   await  svctm  %util
sda2              0.00  1242.00    0.00  133.50     0.00 11008.00    82.46     0.02    0.17   0.11   1.45

The Virtual drive I/O usage that is high during the same time:-

Device:         rrqm/s   wrqm/s     r/s     w/s   rsec/s   wsec/s avgrq-sz avgqu-sz   await  svctm  %util
drbd10            0.00     0.00    0.00   23.50     0.00   188.00     8.00     3.41    6.11  38.17  89.70

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
          24.42    0.00    0.69    0.06    0.00   74.83

For checking the I/O usage for DRBD, which drive should we be monitoring??

On googling i found this link:- http://lists.linbit.com/pipermail/drbd-user/2012-January/017622.html

Manu
  • 101
  • 6
  • seems like a duplicate of: http://serverfault.com/questions/807809/i-o-high-on-drbd-disk-drbd10-on-stacked-site/808003#808003 – Matt Kereczman Oct 18 '16 at 20:07
  • Not really. That question was for tuning the performance for drbd, and this one is to get more information and confirmation on a known issue with drbd reporting of I/O. Please treat both as separate question. – Manu Oct 19 '16 at 07:44

1 Answers1

1

It has been known for awhile that the iostats on the virtual DRBD device have been wildly inaccurate. It is suggested to just simply ignore them.

Instead simply check the iostats of the backing disk which DRBD is configured to use. This should give you an accurate idea of the IO.

Dok
  • 1,110
  • 1
  • 7
  • 13
  • Thanks @Dok, I am sorry,, i did not find this anywhere in the drbd documentations or as known defect. If there is any reference you are aware, please guide me to that. – Manu Oct 18 '16 at 15:45
  • It is not found in any official documentation that I know of, but the topic has come up a few times on the mailing list. One such reference would be the one you have already linked to. Another can be found here: http://lists.linbit.com/pipermail/drbd-user/2012-April/018326.html – Dok Oct 18 '16 at 17:58
  • Thanks @Dok. Is there any version upgrade of drbd available for which this has been fixed and released, of which you are aware. – Manu Oct 19 '16 at 07:49
  • I really do not know if it has been "fixed" in an update. Just simply look at the statistics of the backing disk. These will be accurate. – Dok Oct 25 '16 at 21:17