On 2009-05-08 09:53, John Lange wrote:

Being the only "*nix" person in our office, I've been designated to fix a problem on a old SCO server.

I know what you're thinking because I thought the same thing; "it's a dream come true!" but I didn't let my enthusiasm show.

The problem is that cron was "hanging" in the middle of the night roughly every second day. The log show that it was throwing an error: "Cannot create temporary file" around 4:30am every second day, but the exact time and process running isn't always the same.

/tmp and /usr/tmp are normal looking and there should be enough free disk space so that isn't the problem.

I did a Google search on your subject line, and the only suggested solution seems to be what you've already considered.

Thinking that maybe something was suddenly chewing up disk and then being cleared I setup a little cron to log disk usage every 15 minutes.

Now cron hangs _every day_ at about 16:30!

Am I correct in assuming that the added crontab entry increased the frequency of occurrence?

In order to restart cron, its FIFO pipe file has to be deleted and the box rebooted.

Again, disk usage is not a problem.

Anyone know anything about how SCO cron works?

I know nothing specific about SCO cron, but...

Given that the frequency of occurrence seems to have increased after you added new cron entries, I'm thinking it's a process or system resource allocation issue. Check your per-process open file limit as well as system-wide open file limit. Also, check any other limits that might affect file creation/opening.

Early UNIX systems had hard-coded limits for a lot of in-kernel tables, such as open file table, cached inode table (yes, that was separate), etc. When you wanted to tweak these, you'd have to rebuild your kernel. Not sure how SCO does things, nor what specific limits there would be, how you'd change them, etc.