how does the VV go into the undefined PD?

[dennischen]

Hi Everyone,

I have a question about the vv space allocation. The storage is 3par8400. It only has some SSD disks and 15K rpm SAS disks. Recently I add 6 10K rpm SAS disks in the ss8400, and I didn't do the tunesys job. Here is the question, I found the chunklets in the 10K SAS disks have been used.

SS8400-2 cli% showpd -c -p -rpm 10
-------- Normal Chunklets -------- ---- Spare Chunklets ----
- Used -- -------- Unused -------- - Used - ---- Unused ----
Id CagePos Type State Total OK Fail Free Uninit Unavail Fail OK Fail Free Uninit Fail
60 0:22:0 FC normal 1116 202 0 722 0 0 0 0 0 192 0 0
61 0:23:0 FC normal 1116 175 0 750 0 0 0 0 0 191 0 0
64 1:22:0 FC normal 1116 199 0 726 0 0 0 0 0 191 0 0
69 1:23:0 FC normal 1116 166 0 759 0 0 0 0 0 191 0 0
70 2:22:0 FC normal 1116 202 0 723 0 0 0 0 0 191 0 0
71 2:23:0 FC normal 1116 183 0 742 0 0 0 0 0 191 0 0
-----------------------------------------------------------------------------------------
6 total 6696 1127 0 4422 0 0 0 0 0 1147 0 0

So, I check one of the pd. Found there are VVs in the pd.

SS8400-2 cli% showpdvv 60
PDId CagePos Type RPM VVId VVName VVSp
60 0:22:0 FC 10 1 .srdata usr
60 0:22:0 FC 10 28 copyvmwaredata1_5T usr
60 0:22:0 FC 10 28 copyvmwaredata1_5T adm
60 0:22:0 FC 10 28 copyvmwaredata1_5T snp
60 0:22:0 FC 10 84 copyvmwaredata2_5T usr
60 0:22:0 FC 10 84 copyvmwaredata2_5T adm
60 0:22:0 FC 10 84 copyvmwaredata2_5T snp

Here is the vv and cpg information. These VVs are in the FC_r5 CPG.

SS8400-2 cli% showvvcpg
-------(MB)-------- -New (MB)--
Name CPG Adm Snp Usr Adm Snp Usr
admin --- 0 0 10240 0 0 0
.srdata --- 0 0 81920 0 0 0
copyvmwaredata1_5T FC_r5 640 16256 5242880 0 0 0
copyvmwaredata2_5T FC_r5 512 16256 5242880 0 0 0
copyvmwareoracle_1.5T SSD_r5 1152 16256 1536000 0 0 0

But the CPG in defined with only 15K rpm disks when creating. So how does these VVs use the undefined 10k PD?

SS8400-2 cli% showcpg -sdg
------(MB)------
Id Name Warn Limit Grow Args
0 FC_r1 - - 32768 -p -devtype FC -rpm 15
1 FC_r5 - - 32768 -t r5 -ha mag -ssz 5 -ss 128 -p -devtype FC -rpm 15
2 FC_r6 - - 32768 -t r6 -ssz 6 -ss 128 -p -devtype FC -rpm 15
3 SSD_r1 - - 8192 -ss 32 -p -devtype SSD -rpm 150
4 SSD_r5 - - 8192 -t r5 -ha mag -ssz 5 -ss 64 -p -devtype SSD -rpm 150
5 SSD_r6 - - 8192 -t r6 -ssz 6 -ss 128 -p -devtype SSD -rpm 150

[yapkl]

Hi dennischen,

This is a 3PAR 8400 2N correct?
What is the 3PAR OS version?
What is the available free space on the 15K SAS PDs?

[dennischen]

Hey,

Yes, it is a two node 8400. I cannot get the OS version now....But the 15K SAS PDs are full, no free space..  

So when the PDs has no free space. The VVs will allocate the another PDs? how is rules. What if there is some SATA disks availale.

[Dennis Handly · Accepted solution]

> So when the PDs has no free space. The VVs will allocate the another PDs?

Yes, these are called desparation grows and it will pick other devtypes, etc.

> What if there is some SATA disks availale.

No SATAs, just SAS.  Did you mean NL?  Yes, it will finally chose them too.

[dennischen]

Thanks for the help. 

[dennischen]

BTW, do you know any official guides or white paper, that has the descirption of desparation grow? I can provide it to the customer...

[Dennis Handly]

@dennischen:  do you know any official guides or white paper

No, I did find a mention of a fix in 3.1.3 MU3 RN

73521  The compactcpg command now does not move data into non-matching LDs.

If there are LDs that do not match the CPG's growth characteristics, most likely generated due to desperation grows, then do not move data from LDs that match the CPG's growth characteristics when performing a compactcpg.

Any alerts saying it did a desperation grow?