Infrastructure Software

Hi,

I am looking for some assistance in interpreting some performance statistics from a recent production issue.

We are running Solaris 9 on SunFire v490 with 8 CPUs, I think (I'm not the sys admin, so I'm not 100% on these details - its definitely Solaris 9, but the rest might be slightly off!)

The issue was that durning processing of a particularly mailshot (using sendmail) the load averages for the box climbed up to about 80. At this stage average CPU on the box was only about 50% - higher than normal but not massive. There were a lot of processes (~3,000) created by the mail job.

I wondering how to interpret these load average figures, in light of the CPU and num processess? Was the box actually stressed out during this time? The box in question also hosts the primary database for a major web system, but usage at the time of the mail shot was low (it was early am). However, we are concerned that if mailshot had run a bit longer, it might overlap with increased web user activity and impact site responsiveness for our customers.

Any thoughts re. this would be greatly appreciated.

Regards,

Adrian