CPG auto growth failed, its RawFree & LDFree is 0 MB

[louis ji]

Hi,
There is one 8200, 3.2.2 (MU4). It has 8*1.2T SAS drives and 12*4TB NL drives.
Only one CPG uses the NL drives, named NL_r5.
Last weekend, I add 4*1.2T SAS drives and 4*4TB NL drives. In that time the LDFree of NL_r5 cpg is around 3.7 TB.
I run admithw without -notune option.

Yesterday, I found shows two alters:

ThrType : CPG was grown with degraded parameters
Component : sw_cp:3:NL_r5
Message : CPG 3 (NL_r5) could not grow with its normal grow parameters. The following parameters were used: createald -wait 0 -cpsd NL_r5 -t r5 -ha mag -ssz 6 -ss 128 -ch first -p -devtype NL -n tp-3-sd-1 -sz 8192


Type : CPG growth failure
Component : sw_cp:3:NL_r5
Message : CPG NL_r5 SD and/or user space grow command: createald -wait 0 -cpsd NL_r5 -t r5 -ha mag -ssz 6 -ss 128 -ch first -p -devtype NL -n tp-3-sd-2 -sz 8192 Failed with: The layout for volume size 8192 MB, RAID 5, set size 6, devtype NL, availability "ha mag" requires 6 magazines connected between a nodepair but this system only has a maximum of 4 magazines connected between a nodepair.

And the RawFree and LDFree value under EstFree of NL_r5 cpg is 0 MB.

I run tunesys manually, but showtask -d <tunesys-id> shows 0 tunes finished.

What should I do in this status?

[MammaGutt]

Hi,
There is one 8200, 3.2.2 (MU4). It has 8*1.2T SAS drives and 12*4TB NL drives.
Only one CPG uses the NL drives, named NL_r5.
Last weekend, I add 4*1.2T SAS drives and 4*4TB NL drives. In that time the LDFree of NL_r5 cpg is around 3.7 TB.
I run admithw without -notune option.

Yesterday, I found shows two alters:

ThrType : CPG was grown with degraded parameters
Component : sw_cp:3:NL_r5
Message : CPG 3 (NL_r5) could not grow with its normal grow parameters. The following parameters were used: createald -wait 0 -cpsd NL_r5 -t r5 -ha mag -ssz 6 -ss 128 -ch first -p -devtype NL -n tp-3-sd-1 -sz 8192


Type : CPG growth failure
Component : sw_cp:3:NL_r5
Message : CPG NL_r5 SD and/or user space grow command: createald -wait 0 -cpsd NL_r5 -t r5 -ha mag -ssz 6 -ss 128 -ch first -p -devtype NL -n tp-3-sd-2 -sz 8192 Failed with: The layout for volume size 8192 MB, RAID 5, set size 6, devtype NL, availability "ha mag" requires 6 magazines connected between a nodepair but this system only has a maximum of 4 magazines connected between a nodepair.

And the RawFree and LDFree value under EstFree of NL_r5 cpg is 0 MB.

I run tunesys manually, but showtask -d <tunesys-id> shows 0 tunes finished.

What should I do in this status?

You should find a time machine and go back in time to last weekend and run tunesys.

If you do showtask it should show the current status now. As it seems that some/most PDs are full it will start to move chunklets (slow) from the fullest PDs to allow future growth as fast as possible before doing a faster tune later (but does require some free space and treats every PD equal).

Then you should also upgrade 3PAR OS to 3.3.1 MU5 which in addition to fixing a lot of issues also includes Express Layout for FC and NL allowing both nodes in the system to access all drives in the system as their primary path. That should also allow you to change to R6 which has generally been the industry standard the last 5 years or so, as the risk of data loss with single parity is not longer worth it with the increased drive size.

For 3PAR, R6 for NL has actually been the recommendation for as long as I've worked with 3PAR (9 years).