3PAR OS 3.3.1 MU1 is released

[mitchellm3]

We just got done our 4th array upgrade last night. Two 7440's last week and two 20840s this week. The upgrade was smooth. Now, the unfortunate thing was while we were upgrading this gem came out...

https://support.hpe.com/hpsc/doc/public ... 1FDF2255C/

Apparently HPE still doesn't have the metadata issue corrected with the DECO volumes. Now snapshots and replication are affected. HPE said they don't see this a lot BUT...you guessed it...it should be fixed in MU2.

Our plan is still to upgrade the rest of our 3pars off of 3.2.2 to 3.3.1 emu1

[bajorgensen]

This is another issue than the previous issue I think.
Seems like you should be good as long as you hold off compression until MU2.

- Anders

[smccreadie]

This is getting really old.
We recently purchased a 9450 array. When we bought it we calculated TCO based on using compression and dedupe. This array was expensive (and its a very high performance so far, worth the cash), so naturally mgmt wants it in production as soon as possible.
Its frustrating to watch HPE fumble this one so hard.
Does anyone have any idea when we could see MU2?

Thanks

[storsnapper]

This is getting really old.
We recently purchased a 9450 array. When we bought it we calculated TCO based on using compression and dedupe. This array was expensive (and its a very high performance so far, worth the cash), so naturally mgmt wants it in production as soon as possible.
Its frustrating to watch HPE fumble this one so hard.
Does anyone have any idea when we could see MU2?

Thanks

Trust me, you are in similar positions as a few of us. Both 8450 arrays here were purchased with space calculation based on getting 2:1 dedupe/compression and we were unable to complete the migration without purchasing additional disk space. 3.3.1 code has been a cluster so far.

[mitchellm3]

I believe MU2 should be available sometime around the middle of March. We have other fixes in MU2 that we are waiting for.

I have been having the same conversations with our HPE representatives about the TCO factor and that it makes the 3par look more affordable. They are funneling that up the chain and I hope you do the same. We are pushing on them to come out with a fix for the issue before mu2 but who knows if they will be able to do that. Some things require more of a rewrite.

We are also asking for more clarification on how bad the metadata issue really is. With EMU1, there is a script that can fix the metadata issue on the fly if it is experienced. However, sometimes the volume can be taken offline and that will require support to fix the issue. That downtime is what concerns us. Is compression worth that risk? I don't think so.

[smccreadie]

I believe MU2 should be available sometime around the middle of March. We have other fixes in MU2 that we are waiting for.

I have been having the same conversations with our HPE representatives about the TCO factor and that it makes the 3par look more affordable. They are funneling that up the chain and I hope you do the same. We are pushing on them to come out with a fix for the issue before mu2 but who knows if they will be able to do that. Some things require more of a rewrite.

We are also asking for more clarification on how bad the metadata issue really is. With EMU1, there is a script that can fix the metadata issue on the fly if it is experienced. However, sometimes the volume can be taken offline and that will require support to fix the issue. That downtime is what concerns us. Is compression worth that risk? I don't think so.

These are my thoughts exactly.
I escalated this issue with my HPE rep as well, told him that without DECO volumes and the associated data compaction ratio, we cannot meet out TCO goals (that were sold to us as well).
I believe they are going to get back to me this coming week, once I hear I will reply to this thread. I'm hoping they issue a patch just for the metadata issue as soon as possible, before MU2 if possible. I already waited a month for EMU1 to be installed before I began using the array in the first place, more delays are totally unacceptable at this point. Also, I now have to create new non-compression volumes and migrate all my data, only to have to migrate back to DECO again once its patches. Yikes.

If you dont mind, what are some of the other fixes you are waiting on in MU2? I want to know if there is anything else I need to look out for!

Thanks

Sean

[Patrick]

This is getting really old.
We recently purchased a 9450 array. When we bought it we calculated TCO based on using compression and dedupe. This array was expensive (and its a very high performance so far, worth the cash), so naturally mgmt wants it in production as soon as possible.
Its frustrating to watch HPE fumble this one so hard...

Ditto, we are currently in the bidding process of a SAN refresh and our var has been informed not to use any deco numbers for the 3Par's. If the feature isn't stable, it doesn't count.

[Cleanur]

Lookout for P20....

[mitchellm3]

The other issue we are waiting to have fixed in a problem with snapshot updates from physical copies. We have an automated refresh process that happens every night and sometimes when VCs are updated from the PCs, VCs that aren't a part of the update will have connectivity issues. HPE has identified the issue and thinks they'll have it fixed in MU2.

On another note, our SE said they may have the new metadata fix coming out in January instead of waiting for MU2. That might be what Cleanur was referencing.

[storsnapper]

Lookout for P20....

Will be applying patch 20 this weekend. We ran into an issue with the new code over the holiday break and HPE is recommending this patch + power cycling the array.

Component Configuration Lock Hold Time
Event Type e001c
Symptom lock hold seconds: 0, virtual volume lock count: 0, ioctl request count: 9, mcall active count: 9, mcall request waiting count: 0, mcall request blocked count: 0, mcalls (msec/name/pid): 558103/MC_CHECK_CONSISTENCY/43557 5095968938/MC_NEVER_RETURN/21114 12008/MCJB_SHOW/43774 1066/MCCP_ENV/43730 6407/MCLP_FCOE_PORT_INFO/43843 1309/MCLP_SHOW/43577 6400/MCJB_SHOW/43844 15473/MCLP_SHOW/43757 1065/MCLP_FABRIC/43845. (Critical)
Analyzed Thu Dec 28 21:52:46 UTC 2017

Description
lock hld sec: 0, vvol lck ct: 0, ioctl ct: 9, mcall act ct: 9 We would recommend you to restart the 3Par nodes one by one and monitor the unit for any errors and revert.