HPE EVA Storage

EVA6000 unexpected perfomace degradation

 
Andrey Dudarev
New Member

EVA6000 unexpected perfomace degradation

Hi.

I have EVA6000 2c8d.

Yesterday i found great reading perfomance degradation. 50 Gb backup went 10 hours. Reading speed is now about 10 Mb/sec, write speed is more then 100. I don't do anything with EVA and servers. Degradation is absolutely unexpected.

How can i find the problem?
5 REPLIES 5
Johan Guldmyr
Honored Contributor

Re: EVA6000 unexpected perfomace degradation

Hey,

check Command View - any issues reported?
check EVA controller eventlog - anything going on there? Maybe ask HP for help to go through the logs?

Are you only having problems with the backup-server or everything else too?

check the SAN - maybe there is something bad going on there (an hba, eva host ports, sfp, cable suddenly broke maybe).
Andrey Dudarev
New Member

Re: EVA6000 unexpected perfomace degradation

Hi

In CommndView everything OK. Fll operation statuses are good. Event log does not contain anything interesting. Only ordinary events.
I expecting problems with all servers (HP-UX, Windows, VMWare).
Hardware seems to be OK. All SAN ports are good.
Today I try to change path on Windows server using MPIO to another controller. Speed does not increese.

Only idea is to power off one controller and look at speed changes.
Johan Guldmyr
Honored Contributor

Re: EVA6000 unexpected perfomace degradation

Hi, yeah that would be worth a shot. It would tell you that something relating to that controller might be broken. But it may still be cable, sfp, port in controller or switch.

I would start looking at port errors on the SAN-switch ports. Maybe they all look good but there is a port flapping or sending bad frames..
Andrey Dudarev
New Member

Re: EVA6000 unexpected perfomace degradation

I look at SAN switches and does not seen any problems. Can you tell me where can see port problems? Which counters can i watch at?
Johan Guldmyr
Honored Contributor

Re: EVA6000 unexpected perfomace degradation

Command: porterrshow

beware that they are historical (if you never cleared them, they are from the Beginning).

statsclear / portstatsclear might be useful to you.

Some examples of counters:
crc errors, encoding errors, LLI (low level interface errors, a collection of other counters but low_level so transmit/light etc.)

If it is increasing a lot on only one port, then maybe that's where your problem is. But if the errors is increasing on several ports then you need to do a little elimination ;)