A few weeks ago, I wrote about the Application Runtime Expert – Network Health Checker. IBM recently made another utility available via the Application Runtime Expert that helps you identify damaged objects.
Objects can get damaged for a variety of reasons and you may not know you have damaged objects until you actually access the object (or the part of the object) that is damaged. One of the more common reasons for damage is a disk failure of some sort – a disk controller failure for example; in the case of disk failures, some of the data may not have been correctly written to disk resulting in damage.
You need to identify damaged objects so you don’t lose data; the typical way to do this has been to perform a full system save. However, if you’re not sure whether you have damaged objects, you have to save everything to cause the system to read all parts of all objects. However, there is a catch-22: if the save encounters a damaged object, the save itself will fail, thus making it difficult to identify all the damaged objects. The damage detection tool can make this process easier.
The use of this utility is very similar to that of the Network Health Checker that I wrote about previously. There is a new parameter on the areVerify script. To run the damaged object detection tool, you go into QShell and use the following incantation (note that you need to have *ALLOBJ special authority):
/QIBM/ProdData/OS/OSGi/templates/bin/areVerify.sh –storage diskUnits=*ALL
You can also request a check for damage only on specific disk units. For example:
/QIBM/ProdData/OS/OSGi/templates/bin/areVerify.sh -storage diskUnits=1,2,4
While the damage detection utility is running, you will find several jobs get submitted to the QCTL job queue. In my example, they run under my user profile since I initiated the utility. You will see something like the following for the jobs that are submitted:
The jobs run at a priority such that normal work on the system should not be affected. However, because the damage detection tool is reading everything on disk, you will notice increased I/O activity.
A report is generated with the results of the check – it is stored in /tmp and is named areDodReport.txt. If damage is detected, the report will include information about the damaged object(s). Of course, my test system was uninteresting and I had no damaged objects.
Most of you are probably more comfortable looking at this via WRKLNK, as the screen capture below shows.
As I discussed earlier in this blog, this utility identifies one type of damage – it’s called a data check. At this time, it will not identify all types of damage for you since there are many reasons damage can occur. However, if you’ve had a disk failure, this utility is a great approach to check for damage.
The article Automate Damaged Object Detection on Your System, written by Tim Rowe, provides more details that you may find useful.
This enhancement is available on both 7.1 and 6.1. It was delivered via PTFs. The information below lists the PTFs that you will need. Order the PTF in bold and include the requisites so you will get all the fixes you need in one order.
- 7.1 - SI50374 (5770SS1)
- Make sure the following requisite are also applied:
For 7.1, the HTTP Server group PTF (SF99368) includes these PTFs; if you install at least level 21 of this group, you will have the above PTFs.
- 6.1 - SI45499 (5761SS1)
- Make sure the following requisites are also applied:
- SI51025 (5761SS1)
- SI30796 (5761SS1)
- MF57435 (5761999)
- MF57436 (5761999)
- MF57425 (5761999)
- MF57426 (5761999)