vBulletin Search Engine Optimization
| |||||||
| Register | FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| ||||
| Hi all having a disk configuration like that hdisk1 None None hdisk2 None None hdisk3 None None hdisk4 None None .... and so on .... and hdiskpower1 PVID vgname1 hdiskpower2 PVID vgname1 .... and so on .... With powermt I see every hdiskpower contains 4 hdisks - ok. Now I add 16 new luns and after emc_cfgmgr and powermt config I see 16 new hdisks and 4 new hdiskpower - ok. hdiskpower3 PVID None hdiskpower4 PVID None hdiskpower5 PVID None hdiskpower6 PVID None Every filesystem which I want to increase is on two hdiskpower, say lvname fsname 404 PPs hdiskpower1 lvname fsname 066 PPs hdiskpower2 After increasing the filesystems, "old" hdiskpower should be removed, means I want to migrate the part of a fs, which is actually on hdiskpower2 to new hdiskpower3. So I plan to do the following: extendvg vgname1 hdiskpower3 migratepv hdiskpower2 hdiskpower3 check if hdiskpower2 is empty after migrating if so, reducevg vgname1 hdiskpower2 and remove the hdiskpower2 with tools of powerpath Can this be done this way? Thx in advance Friedhelm |
| |||
| On 2008-03-04, Friedhelm Neyer <[email protected]> wrote: > extendvg vgname1 hdiskpower3 > migratepv hdiskpower2 hdiskpower3 > check if hdiskpower2 is empty after migrating > if so, reducevg vgname1 hdiskpower2 > > and remove the hdiskpower2 with tools of powerpath > > Can this be done this way? Yes, this will work quite nicely. Before you remove hdiskpower2, don't forget to make a note about which hdisks make up hdiskpower2. After you remove hdiskpower2, you should remove the hdisks too. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad. |
| |||
| On Mar 4, 6:49 pm, Friedhelm Neyer <[email protected]> wrote: ... > and remove the hdiskpower2 with tools of powerpath > > Can this be done this way? Normaly no need to use the emc tools. An rmdev work fine. But you should read .... Crash in scsidisk_process_sense when using non-MPIO disks. http://www14.software.ibm.com/webapp...D=4119#IZ14308 .... which happens also on AIX 5.3. It means more less: First remove the hdiskpower and the depending hdisk with rmdev BEFORE you remove the lun from the storage group. Otherwise a simple lspv might crash your system ... cheers Hajo |
| |||
| > Normaly no need to use the emc tools. An rmdev work fine. more often than not, this will work fine. but if you ever have any problems with emc devices, emc will tell you that you should have followed their documented procedures. following their procedures is much easier than the problem resolution after the fact. |
| |||
| Started the first migartepv this morning - 4 log. volumes with about 130 GBm it runs round about 3 h and all gone fine. Friedhelm Neyer schrieb: > Hi all > > having a disk configuration like that > > hdisk1 None None > hdisk2 None None > hdisk3 None None > hdisk4 None None > ... > and so on > ... > > and > hdiskpower1 PVID vgname1 > hdiskpower2 PVID vgname1 > ... > and so on > ... > > With powermt I see every hdiskpower contains 4 hdisks - ok. > > Now I add 16 new luns and after emc_cfgmgr and powermt config I see 16 > new hdisks and 4 new hdiskpower - ok. > > hdiskpower3 PVID None > hdiskpower4 PVID None > hdiskpower5 PVID None > hdiskpower6 PVID None > > Every filesystem which I want to increase is on two hdiskpower, say > > lvname fsname 404 PPs hdiskpower1 > lvname fsname 066 PPs hdiskpower2 > > After increasing the filesystems, "old" hdiskpower should be removed, > means I want to migrate the part of a fs, which is actually on > hdiskpower2 to new hdiskpower3. So I plan to do the following: > > extendvg vgname1 hdiskpower3 > migratepv hdiskpower2 hdiskpower3 > check if hdiskpower2 is empty after migrating > if so, reducevg vgname1 hdiskpower2 > > and remove the hdiskpower2 with tools of powerpath > > Can this be done this way? > > Thx in advance > Friedhelm > |
| |||
| Hajo Ehlers wrote: > On Mar 4, 6:49 pm, Friedhelm Neyer <[email protected]> > wrote: > .. >> and remove the hdiskpower2 with tools of powerpath >> >> Can this be done this way? > Normaly no need to use the emc tools. An rmdev work fine. But you > should read think twice about that. depending on which version of ppath you use you risk odm corruption (or at least having emc claim odm corruption when something barfs and the resulting cleanup requires a reboot). the best practice, per most of the emc docs spanning the versions we've had to mess with is.. inq | tee /tmp/inq.out # nice to just have this list beforehand powermt remove dev=<power device to remove> rmdev -dl <power device to remove> rmdev -dl <each hdisk for the power device to remove> -r |
| |||
| On 2008-03-09, No Body <[email protected]> wrote: >> Normaly no need to use the emc tools. An rmdev work fine. But you >> should read > > think twice about that. depending on which version of ppath you use you > risk odm corruption (or at least having emc claim odm corruption when > something barfs and the resulting cleanup requires a reboot). the best > practice, per most of the emc docs spanning the versions we've had to > mess with is.. > > inq | tee /tmp/inq.out # nice to just have this list beforehand > powermt remove dev=<power device to remove> > rmdev -dl <power device to remove> > rmdev -dl <each hdisk for the power device to remove> There are at least several Powerlink articles which say you can remove powerpath devices with rmdev, without using powermt remove. I didn't search any other documentation (such as whitepapers, etc) but I expect that rmdev pops up there too. Note that the documentation does state you need to do a "powermt save" after a powerpath reconfiguration. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad. |
| ||||
| No Body schrieb: > Hajo Ehlers wrote: >> On Mar 4, 6:49 pm, Friedhelm Neyer <[email protected]> >> wrote: >> .. >>> and remove the hdiskpower2 with tools of powerpath >>> >>> Can this be done this way? >> Normaly no need to use the emc tools. An rmdev work fine. But you >> should read > > think twice about that. depending on which version of ppath you use you > risk odm corruption (or at least having emc claim odm corruption when > something barfs and the resulting cleanup requires a reboot). the best > practice, per most of the emc docs spanning the versions we've had to > mess with is.. > > inq | tee /tmp/inq.out # nice to just have this list beforehand > powermt remove dev=<power device to remove> > rmdev -dl <power device to remove> > rmdev -dl <each hdisk for the power device to remove> > > -r So I've done the following steps: powermt remove dev=<hiskpowerxx> rmdev -Rdl every hdiskpower which should be removed rmdev -Rdl every underlying hdisk powermt save .... and all runs fine |