As a followup to my
previous post, I now have my database server running on
Solaris Express svn_28. It has been running for a little over a week now.
I was not able to use my original disk partitioning scheme. One of the 18GB drives would not disklabel. I suspect that it has odd firmware since it originally came from an HP disk array. The other, otherwise identical, 18GB drive labeled without any trouble.
I decided to just pull the 18GB drives and use them on a OpenBSD system since they have worked under OpenBSD in the past.
I ended up installing the OS on a 8GB slice on one of the 73GB drives. The rest of drive 0 and drives 1 through 3 are part of a ZFS pool running RAID-Z. Unfortunately, that leaves my OS volume rather exposed.
Yes, I have a ZFS pool made up of a partition on one drive and 3 whole drives. I assume that I'm incurring a performance penalty for mixing it up, but it hasn't caused any problems yet. I cannot recall reading that you are allowed to create uneven RAID-Z pools, so I was pleasantly surprised.
Postgres is currently running in a zone. It (postgres) is being managed by SMF. No problems with Postgres yet, but I'm going to be throwing 2GB of data into it later tonight. I'm looking forward to seeing what that does to the system. Postgres is using a filesystem on the uneven RAID-Z pool for storage.
I finally plugged in and reset the RSC card. Also, I'm using IPF on the global zone to filter traffic for all the zones.
And that is all that is going on with my experimental database server.