CPG failed caused by copy space grow failed

[dennischen]

Hi,

I have a 3par 8400 storage, A CPG was failed caused by  the lack of space to grow. In the IMC and SSMC, I can see the CPG NL_r6 was failed.

Then, I deleted some snapshot in the NL_r6 CPG, and compact the NL_r6 CPG, in order to get some more space. Waiting for the free chunklet growing, but the CPG is still failed... 

Someone suggests the setsysmgr restart might work... Is it necessary restart the sysmgr process to fix this problem? and NOT sure is there any risk while doing the sysmgr restart in production.. Thanks for any advise.

SS8400-2 cli% showpd -c -p -devtype NL
-------- Normal Chunklets --------- ---- Spare Chunklets ----
-- Used -- -------- Unused -------- - Used - ---- Unused ----
Id CagePos Type State Total OK Fail Free Uninit Unavail Fail OK Fail Free Uninit Fail
72 3:0:0 NL normal 3611 3089 0 77 25 0 0 0 0 420 0 0
73 3:1:0 NL normal 3611 2997 0 40 154 0 0 0 0 420 0 0
74 3:4:0 NL normal 3611 2997 0 42 152 0 0 0 0 420 0 0
75 3:5:0 NL normal 3611 3089 0 77 25 0 0 0 0 420 0 0
76 3:8:0 NL normal 3611 3089 0 77 25 0 0 0 0 420 0 0
77 3:9:0 NL normal 3611 2997 0 41 153 0 0 0 0 420 0 0
78 3:12:0 NL normal 3611 2997 0 42 152 0 0 0 0 420 0 0
79 3:13:0 NL normal 3611 3089 0 77 25 0 0 0 0 420 0 0
80 3:16:0 NL normal 3611 3089 0 77 25 0 0 0 0 420 0 0
81 3:17:0 NL normal 3611 2997 0 42 152 0 0 0 0 420 0 0
82 3:20:0 NL normal 3611 2997 0 43 152 0 0 0 0 419 0 0
83 3:21:0 NL normal 3611 3089 0 78 25 0 0 0 0 419 0 0
------------------------------------------------------------------------------------------
12 total 43332 36516 0 713 1065 0 0 0 0 5038 0 0

[Sheldon Smith]

The CPG is failed because you've run out of free NL chunklets.
Messing around with system processes (not a Good Idea in any event) is not going to make physical storage magically appear. You need to wait for those uninitialized chunklets to get reclaimed.
At which time it looks like you will have around 1700 chunklets (GiB), free. Which will still be less than the 24 GiB requested mentioned in the details of alert 143.
Delete more snapshots in the NL_r6 CPG; maybe you have enough to clean out and make room. More likely, it's time to buy more NL drives to add to the system.

[Dennis Handly]

> A CPG was failed caused by the lack of space to grow.

You may want to enable the growth warning for that CPG.

[dennischen]

yes, All the snapshots are deleted. And looks like I will get 1700GiB free space. I believe this is enough for the CPG growth. Why is it still less than the 24GiB?

Thanks.

[Sheldon Smith]

Oops. I wrote that before coffee and forgot to mentally shift a decimal point. :}
By the way, I'd recommend you start switching over to the browser-based SSMC interface.

As close as the system is to full, I'd suggest editing the CPG growth increment down to 8G.
It looks like your AO configuration includes a NL RAID6 CPG which would be competing for those free NL chunklets.

Are the NL drives's chunklets initialized, or are a few still in progress?
When all the chunklets are re-initialized, try the allocation again. If, once everything is initialized you still are getting a failure on the CPG, contact 3PAR Support.

Looking at the CPGs above, over half of your allocated storage is in NL drives. You might consider adding some which would also (presumably) off-load some of that FC storage.