Case Study: Configuring and Monitoring for Mirrored Disks

I l @ ve RuBoard

At Silicon Valley Hospital, Nancy Jergens is responsible for maintaining patient records on the hospital's records computer system. To provide extra data protection and high availability, she has decided to use MirrorDisk/UX to mirror the /patient filesystem containing the patient records.

Nancy sets up two-way mirroring on the logical volume containing the patient records. The logical volume lvol_patients is mounted at /patient.

Verifying Configuration

After Nancy configures the mirroring, she verifies the configuration by using the lvdisplay command. As shown in Listing 5-12, Nancy verifies that the logical volume, lvol_patients, is available and mirrored with one copy.

Setting Up Monitoring

Nancy knows that if one of her mirrored copies fails, her patient data will be protected, because she has another copy. But, she wants to be told when she has only one copy left. Unless she monitors for this, she won't be told when a mirrored copy fails.

Listing 5-12 Using lvdisplay to verify mirroring status and copies.
 # lvdisplay /dev/vg_hospital/lvol_patients - Logical volumes - LV Name                    /dev/vg_hospital/lvol_patients VG Name                    /dev/vg_hospital LV Permission              read/write LV Status                  available/syncd Mirror copies              1 Consistency Recovery       MWC Schedule                   parallel LV Size (Mbytes)           100 Current LE                 25 Allocated PE               50 Stripes                    0 Stripe Size (Kbytes)       0 Bad block                  on Allocation                 strict 

Using the EMS Configuration GUI, Nancy configures EMS DVM to monitor the mirrored copies of the logical volume containing the patient records. As shown in Figure 5-10, she sets up the monitoring request so that she will be notified via SNMP at her OpenView IT/O management station when the number of mirrored copies of the lvol_patients logical volume drops below two. Nancy configures the notification event to have Major severity.

Figure 5-10. Using EMS to monitor mirrored copies.

graphics/05fig10.gif

You may notice that lvdisplay shows 1 as the number of mirror copies, but EMS shows the current number of copies as 2. EMS DVM looks at the total number of copies. lvdisplay reports the number of mirrored copies.

Mirror Fails

Something happens on one of the disks that keeps the data mirrored. MirrorDisk/UX can no longer mirror the data. The data is no longer protected from failure. Fortunately, Nancy is monitoring for this, and an event is sent to the IT/O Message Browser. As shown in Figure 5-11, a major (Maj) severity EMS event is displayed in the Message Browser.

Figure 5-11. IT/O Message Browser shows an EMS event indicating a mirrored disk failure.

graphics/05fig11.gif

Restoring Mirrors

After running the diagnostic tools in STM, Nancy identifies the failed component. After swapping in a new drive, she synchronizes the copies so that each copy of the data is identical.

Verifying Configuration

Now that Nancy has fixed the failed disk, she uses the lvdisplay command to verify that the logical volume is once again available and that she has a mirrored copy of the data.

Nancy continues to monitor the copies by using EMS DVM, knowing that she will be notified if the mirror fails again.

I l @ ve RuBoard


UNIX Fault Management. A Guide for System Administrators
UNIX Fault Management: A Guide for System Administrators
ISBN: 013026525X
EAN: 2147483647
Year: 1999
Pages: 90

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net