IN THE TRENCHES

Routine Purging Will Stop ABENDs

Not long ago, the owner called to report that his server had started ABENDing every night when the backup started. This backup program had been running without a hitch for six months.

Before starting my own business, I was employed by Hewlett-Packard Co. as a customer engineer working on minicomputer systems and networks. As preparation for my job, HP had me attend a trouble-shooting class called Kepner-Tregoe. One of the major tenets of K-T trouble-shooting is to "find out what has changed," and that invariably will be the source of the problem.

The next day, I arrived onsite and discussed the situation with the owner's son. I asked whether he had made any changes, anticipating that he would say yes. Instead, he emphatically said no. I then verified that the server had ABENDed when the backup program started and that the backup NLM was running at the time of the ABEND.

I looked inside the server and verified that the hardware setup was OK. I have a policy of keeping my customers' servers up-to-date with patches and updates, so I verified that the server was updated.

I logged in as Supervisor to check the server and volume error logs and found nothing. One habit I have when I log in to one of my customers' servers is to execute the chkvol command to see how the volumes are fixed for disk space and whether there are a lot of deleted files to be purged. When I ran chkvol, I noticed that the customer had not followed my advice to use the purge /all command to clean out the deleted files. I ran purge /all and again explained to the customer that he should regularly purge deleted files.

I headed back to my office unhappy that I had not found the cause for the ABENDs. I checked the backup vendor's ftp (File Transfer Protocol) site for possible updates but didn't find anything.

The next day, I was pleasantly surprised to learn that the backup had run successfully without ABENDing the server. We discovered after several days that the problem was gone. The only thing I had done was to purge all the deleted files.

My theory is that the backup program needed disk space to create temporary files and that NetWare could not purge the deleted files quickly enough to supply the needed space. This caused the backup NLM to ABEND the server. That was the "change" that created the problem.

Fortunately, the customer has learned that regular purges are a good thing.

Rick Troha, Master CNE and GroupWare CNE, owns Network Support LLC in North Olmsted, Ohio. He can be reached at rickt[at]nwsup.com.

Call for Trenches