I will use rootvg as an example with hdisk0 and hdisk1, where hdisk1 will be the failing one. This manual suppose that failing disk hdisk1 is still available and not in missing state and is hot swap-able and secondary dump device is used on hdisk1.

1. Unmirror rootvg
2. Check where your primary/secondary sysdump logical volume goes
3. If it goes to lv which is on failing hdisk1 redirect it for time being to /dev/sysdumpnull, see sysdumpdev
4. Check if you have any logical volumes left on hdisk1
5. If you have any logical volumes left on failing disk you can move them to hdisk0, if you need to (if you have secondary dump device present you can remove it with rmlv -f secondary_dump_device_lv_name)
6. Reduce hdisk1 from rootvg
7. Take location code of failing hdisk1
8. Set hdisk1 in to define state
9. Use diag to replace and identify failing hdisk1
10. Once done add new replaced hdisk1 to rootvg
11. Create secondary dump device on hdisk1 if needed and point second dump device to it (xx represent the size which should be and it can be calculated with command sysdumpdev -e (in bytes))
12. Mirror rootvg
13. Create boot image on hdisk1
14. Add hdisk1 to bootlist
15. Check bootlist
That should be all. If I forgot something please advice me.

Thank you.