After a crash of the server, during a migratelp operation, one logical volume is in a wrong state, one PP has 3 copie while the other have only 2:
[root@cltest1]/root# lsvg -l RG1vg RG1vg: LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT datalv jfs2 5 10 8 open/syncd /oracle/data1 0516-1147 : Warning - logical volume data1lv may be partially mirrored. data1lv jfs2 1300 2601 9 open/syncd /oracle/data1/data [root@cltest1]/root# lslv -m data1lv | more data1lv:/oracle/data1/data LP PP1 PV1 PP2 PV2 PP3 PV3 0001 0082 hdisk11 0082 hdisk19 0002 0082 hdisk12 0082 hdisk20 .......... 1216 0365 hdisk11 0344 hdisk18 1217 0341 hdisk11 0367 hdisk19 1218 0369 hdisk12 0370 hdisk12 0341 hdisk19 1219 0344 hdisk12 0370 hdisk20 1220 0344 hdisk20 0369 hdisk13 .......
Here you'll find IBM procedure to solve this trouble:
Please find bellow the procedure to remove a PP from an LV: Very Important: MAKE a backup of the VG! Run: 1.Create a mapfile of the PPs: # lquerylv -L `getlvodm -l <lv name>` -r > /tmp/mapfile (The created file contains a listing of the following form: <pvid> <pp> <lp>) . 2. Edit /tmp/mapfile and take out all entries that must NOT! be deleted (i.e. the entries of the complete mirror) 3. unassign the PPs that are listed in /tmp/mapfile # lreducelv -l `getlvodm -l <LV name>` -s <size> /tmp/mapfile (<size> is the number of lines in /tmp/mapfile: # SIZE=`wc -l /tmp/mapfile`) . 4. make an update on the LVCB: # putlvcb -c 1 lvname . 5. # savebase (For saving the changes to the ODM.) Please let me know if you need any assistance or have any questions.
Some usefull commands concerning the subject:
[root@cltest1]/root# synclvodm RG1vg [root@cltest1]/root# syncvg -v RG1vg [root@cltest1]/root# lextendlv -l `getlvodm -l lv01 ` -s 1 /tmp/mapfile [root@cltest1]/root# reorgvg RG1vg