Comments (1)
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

1 frenger commented Permalink

Hi Brian
We could use this but our environment is a bit more complex.

We have 2 productive locations let's call it A and B.
We do not Mirror local VGs like rootvg (because they are on SAN and should be fine) but in our ~ 30 HACMP Clusters we mirror between A and B (this is SAN Storage too). In this environment its crucial to be sure that first, each pp on A is mirrored to B and second, the first PP (lslv -m <lvname>) is the same side where the cluster application primarly runs.
Setting this initally is ok, but AIX is somehow limited when we increase a filesystem by adding additional LUNs. Due we are in a cluster environment all VGs are enhanced concurrent and due that fact the expansion of mirrored filesystems gets a horrible task and so often PPs get often mirrored to the wrong side or second mirror copy on same side what is really worse. In normal vgs the best way to solve this was removing the mirror from site B, expanding the VG only with the Site A LUNs only, increasing the filesystems, then a reorgvg (because most of our lvs have maximum distribution among their disks as intrapolicy setting). When that is finished the LUNs from location B can be added and a mirrorvg can be run. So far the noncomplicated easy way.
In enhanced concurrent VG you can't run a reorgvg and as well as a lot of other chvg commands. But well not all things can be scripted an this may be temporary because I've hreard that they are working on enhanced concurrent vgs to get them at least as good as normal vgs, hopefully somewhere in the future.
But Checking the mirror status to be sure that everything is correctly configured would be nice, but can get complicated due it has to check the serial Number of the Storage disk (lscfg -vl output) an somehow you have to tell him wich serial which site is .... it sounds nice but seems complicated.
Cheers Sebastian