Hmm?

It is is always a quiet time between Christmas and New Year – there are three public holidays in England and a lot of my retail customers are change adverse in their peak trading season so things throttle back. It gives me the chance to get around my teams that are still working (and a lot of my guys, self included, take some leave around now) and just have a chat. Today, I braved the train journey across Central London and visited my team in Epsom. The team provides support for an ETL application that is coming up to retirement, as technology goes, it very much old-school. The hardware is out of vendor support as is the OS, as is the programing language, as is the Oracle database that lurks in the background. There is no clear upgrade path as vendor compatibly matrixes stopped on the versions before the installed versions.

Just before I left them this afternoon a support call came in to advise them that a file system had exceeded the alerting threshold – the file system was the one used for Oracle. A bit of AIX jiggerypokery showed the problem came in a directory of archived Oracle alert logs. It seems that each day the alert log is copied to an archive directory and each week the oldest logs deleted from the archive. And here is the problem, the first lines of the live alert log recorded the database going into backup mode in May 2004 and every day since then it has grown. So a rolling archive of old logs will increase in size and eventually consume the available disk. But what really puzzles me is the need to keep 7 months-worth of log copies;