now being the proud owner of an HDS AMS, I do have a little issue:
I saved the money on the HDLM (loadbalancing multipath) drivers and set for
the (ODM)plugin to regular AIX-MPIO.
This is fine and works as expected, but the attributes default to
'reserve policy=single_path'
Although 'chdev -a etc...' is easy, it is something I do have to remember.
Is there a way to make no_reserve the default policy?
(I am pretty sure I should sink it within the CuDV (or maybe CuAt), but my
last direct interaction with the ODM was on AIX4.3.3 - 8 years ago, so any
hints are welcome.
Thanks
Jan
--
wings and claws
Hi,
it's not in the Cu(stomized)At but the P(re)d(efined)At. Check for:
odmget -q "attribute = reserve_policy and uniquetype like *fcp*mpioosdisk*" PdAt
Should result in something like:
PdAt:
uniquetype = "disk/fcp/mpioosdisk"
attribute = "reserve_policy"
deflt = "single_path"
values = "no_reserve,single_path"
width = ""
type = "R"
generic = "DU"
rep = "sl"
nls_index = 96
If you would put that into an file and edit to:
PdAt:
uniquetype = "disk/fcp/mpioosdisk"
attribute = "reserve_policy"
deflt = "no_reserve" <== !!!
values = "no_reserve,single_path"
width = ""
type = "R"
generic = "DU"
rep = "sl"
nls_index = 96
Then you could change by using "odmchange" or by using "odmdelete" and "odmadd"
the default when cfgmgr is being called.
The steps (I do prefer "delete" and "add" approach)
1) odmdelete -o PdAt q "attribute = reserve_policy and uniquetype like \
*fcp*mpioosdisk*".
Do this only if:
- you saved the text above
- the related "odmget" command results in exactly the entry you want
to delete.
2) Create the file above, let's call it: odmadd.txt
3) odmadd odmadd.txt
DISCLAIMER: It do not take any responsibilty for that, since I do not know your
ODM exactly and I'm not 100% sure, whether the entry above hits exactly the
driver for your disks (May be it is: mpioapdisk)
If you decide to do so, keep in mind: During AIX updates the PdAt may be changed
back again and you need to repeat the procedure. So I would prefer to have a
standard working instruction (and monitoring/regular prove method) to assure
correct settings.
Regards,
Uwe Auer
Hi,
typo.
Step 1) should read:
odmdelete -o PdAt -q "attribute = reserve_policy and uniquetype like \
*fcp*mpioosdisk*"
(missed the "-" in "-q")
Sorry for that.
Regards
Uwe Auer
Ah, certainly. sorry.
[explanation snipped]
Below is what I did on a 5.3TL9SP4 LPAR:
##### note the slashes instead of the stars
##### checking
# odmget -q "attribute = reserve_policy and uniquetype =
disk/fcp/htcAMSmpio" PdAt
PdAt:
uniquetype = "disk/fcp/htcAMSmpio"
attribute = "reserve_policy"
deflt = "single_path"
values = "no_reserve,single_path,PR_shared,PR_exclusive"
width = ""
type = "R"
generic = "DU"
rep = "sl"
nls_index = 96
##### writing odmadd.txt
# odmget -q "attribute = reserve_policy and uniquetype =
disk/fcp/htcAMSmpio" PdAt > odmadd.txt
##### editing odmadd.txt
# vi odmadd.txt
PdAt:
uniquetype = "disk/fcp/htcAMSmpio"
attribute = "reserve_policy"
deflt = "no_reserve" <<<--- changed from 'single_path'
values = "no_reserve,single_path,PR_shared,PR_exclusive"
width = ""
type = "R"
generic = "DU"
rep = "sl"
nls_index = 96
##### *Danger* deleting
# odmdelete -o PdAt -q "attribute = reserve_policy and uniquetype =
disk/fcp/htcAMSmpio"
0518-307 odmdelete: 1 objects deleted.
##### adding
# odmadd odmadd.txt
##### verifying
# odmget -q "attribute = reserve_policy and uniquetype =
disk/fcp/htcAMSmpio" PdAt
PdAt:
uniquetype = "disk/fcp/htcAMSmpio"
attribute = "reserve_policy"
deflt = "no_reserve"
values = "no_reserve,single_path,PR_shared,PR_exclusive"
width = ""
type = "R"
generic = "DU"
rep = "sl"
nls_index = 96
>DISCLAIMER: It do not take any responsibilty for that, since I do not know your
>ODM exactly and I'm not 100% sure, whether the entry above hits exactly the
>driver for your disks (May be it is: mpioapdisk)
Actually, it was 'htcAMSmpio' ;))
>If you decide to do so, keep in mind: During AIX updates the PdAt may be changed
>back again and you need to repeat the procedure.
Since uniquetype is called htcAMSmpio, I expect the next ODM driver package
vom HDS to revert my changes. Until then, it should be save.
But I�ll certainly keep this in mind.
> So I would prefer to have a
>standard working instruction (and monitoring/regular prove method) to assure
>correct settings.
noted.
>Regards,
>Uwe Auer
Thank you very much