AIX LVM in practice: lsvg/lspv/lslv, growing a JFS2 filesystem online, and mirroring rootvg

Goal

Inspect AIX LVM objects, grow a JFS2 filesystem without unmounting it, and mirror the rootvg volume group for disk-failure tolerance.

Prerequisites

Root access; for mirroring, a second disk of adequate size already visible to lspv; for growing a filesystem, free physical partitions in the volume group (check with lsvg <vgname>).

Steps

Inspecting LVM objects

1. lsvg lists volume groups on the system; lsvg <vgname> shows its PP size, free/used partitions and limits (MAX LVs, MAX PVs).
2. lsvg -p <vgname> lists the physical volumes in a volume group and their partition state.
3. lslv -m <lvname> maps a logical volume's logical partitions to the physical volumes and partitions backing them — the AIX equivalent of asking "where does this data actually live."

Growing a JFS2 filesystem online

4. Confirm free space in the volume group with lsvg <vgname>.
5. Increase the filesystem: chfs -a size=+10G /mountpoint grows the mounted JFS2 filesystem without unmounting it. AIX enforces an upper bound: a 4096-byte-block JFS2 filesystem cannot exceed 32 TB, and the operation fails with a message citing the 512-byte-block limit if you try to exceed it.
6. Verify with df -g /mountpoint and lsvg <vgname> to confirm the new free-partition count.

Mirroring rootvg

7. Add the free disk to rootvg, then mirror onto it: extendvg rootvg hdisk1, then mirrorvg rootvg hdisk1 (mirrorvg only uses disks that are already members of the volume group; mklvcopy per logical volume gives finer control but then needs syncvg -v rootvg and chvg -Qn rootvg by hand).
8. Make the new mirrored disk bootable: bosboot -ad /dev/hdisk1 (repeat per additional mirror disk).
9. Add the disk to the boot list: bootlist -m normal hdisk0 hdisk1.
10. On AIX 7.x mirrorvg turns quorum off for rootvg with immediate effect (message 0516-1804); the reboot IBM lists for this step applies only to AIX 5300-06 and earlier. Verify with lsvg -l rootvg that each logical volume shows twice as many PPs as LPs and no stale state.

Expected result

lsvg/lslv/lspv show the intended layout; the grown filesystem reports the new size under df -g; a mirrored rootvg boots from either disk per bootlist -m normal -o.

Limits and test basis

mirrorvg/rootvg mirroring is documented as unsupported on /usr client, diskless and dataless configurations. Growing past the JFS2 maximum filesystem size is a hard limit, not a tunable; plan filesystem sizing with headroom below 32 TB rather than relying on repeated growth.