3par 8200 Cannot dismisspd, "PD is in use"

[Sultan93]

Hello, everyone

3par 8200 3.3.1 MU1, we added new cage with 24 brand new disks installed correctly, but after 2 weeks we did power down procedure of a new cage, and then for the next day after powering up we've got alerts for  12 disk of 24 with status "failed, vacated servicing" after some researching i found in the community the solution for my issue by executing these commands 

servicemag unmark <cage_id>  <pdid>

servicemag clearstatus <cage_id>  <pdid>

removespare <pdid>:a

dismisspd <pdid>

admithw

it helped me to return  7 failed disks of 12 to normal state, and now i have 5 disks still with "failed" state

Servicemag status -d command shows "No servicemag operations logged"

showpdch -from <pdid>  shows "No chunklets information available" for all failed disks

i cannot dismisspd it shows "PD is in use"

this is the result of "showpd -c"

3PAR8200_Karaganda cli% showpd -c
--------- Normal Chunklets ---------- ---- Spare Chunklets -----
-- Used --- -------- Unused --------- - Used - ---- Unused -----
Id CagePos Type State Total OK Fail Free Uninit Unavail Fail OK Fail Free Uninit Fail
48 3:4:0 NL normal       3611 1 0 3466 0 0 0 0 0 145 0 0
49 3:0:0 NL normal       3611 1 0 3466 0 0 0 0 0 145 0 0
50 3:1:0 NL normal        3611 1 0 3465 0 0 0 0 0 145 0 0
51 3:2:0 NL normal        3611 2 0 3465 0 0 0 0 0 144 0 0
52 3:3:0 NL normal        3611 2 0 3465 0 0 0 0 0 144 0 0
53 3:6:0 NL failed          3611 1 0 3466 0 0 0 0 0 144 0 0
54 3:5:0 NL failed          3611 1 0 3466 0 0 0 0 0 144 0 0
55 3:7:0 NL normal        3611 1 0 3466 0 0 0 0 0 144 0 0
56 3:8:0 NL normal       3611 1 0 3466 0 0 0 0 0 144 0 0
57 3:9:0 NL normal       3611 1 0 3466 0 0 0 0 0 144 0 0
58 3:10:0 NL failed        3611 1 0 3466 0 0 0 0 0 144 0 0
59 3:11:0 NL normal      3611 2 0 3465 0 0 0 0 0 144 0 0
60 3:12:0 NL normal     3611 1 0 3467 0 0 0 0 0 144 0 0
61 3:13:0 NL normal      3611 1 0 3467 0 0 0 0 0 144 0 0
62 3:14:0 NL normal     3611 1 0 3466 0 0 0 0 0 144 0 0
63 3:15:0 NL normal     3611 1 0 3466 0 0 0 0 0 144 0 0
64 3:16:0 NL normal     3611 1 0 3466 0 0 0 0 0 144 0 0
65 3:17:0 NL normal      3611 1 0 3466 0 0 0 0 0 144 0 0
66 3:18:0 NL failed        3611 1 0 3466 0 0 0 0 0 144 0 0
67 3:19:0 NL failed        3611 1 0 3466 0 0 0 0 0 144 0 0
68 3:20:0 NL normal    3611 1 0 3467 0 0 0 0 0 144 0 0
69 3:21:0 NL normal     3611 1 0 3467 0 0 0 0 0 144 0 0
70 3:22:0 NL normal     3611 1 0 3467 0 0 0 0 0 144 0 0
71 3:23:0 NL normal      3611 1 0 3466 0 0 0 0 0 144 0 0

As i understand Used chunklets of those "failed" PDs must be 0, and i guess this is the reason that i cannot dismisspd, my questions is: which commands should i perform to successfully  "dismisspd" and return PDs to the "normal" state, and how to move or remove those used chunklets?

if i perform these steps would it help me?

   setpd ldalloc off <pdid > to stop allocation of chunklets on this pd
- movepdtospare -f -vacate <pdid> to reallocate any used chunklets to other disks of same type
- removespare <pdid>:a  to remove any spare chunklet
- dismisspd pdid 

- admithw

[Hhabibi · Accepted solution]

Hello 

if you perform those steps below, you will vacate the pd manually:

   setpd ldalloc off <pdid > to stop allocation of chunklets on this pd
- movepdtospare -f -vacate <pdid> to reallocate any used chunklets to other disks of same type
- removespare <pdid>:a  to remove any spare chunklet
- dismisspd pdid 

- admithw

from what can i see you have a failed chunklet that need to be cleared.

try first to clear the pd error:

showpd -i <PDID>

after that 

clear pd error by the command:

controlpd clearerr <pdwwn>

example below:

Id CagePos State ----Node_WWN---- --MFR-- -----Model------ -Serial- -FW_Rev- Protocol MediaType -----AdmissionTime------
0 0:0:0 normal 5002538A4951AC21 SAMSUNG AREA3840S5xnFTRI 0M504638 3P00 SAS MLC 2019-10-01 15:44:10 EEST

controlpd clearerr 5002538A4951AC21

please try and check.

[Sultan93]

Thank you so much, it helped to return my PDs to a normal state!

[Hhabibi]

hello

you are very welcome. thanks