EVA4400 CA logging state

[WilliamSmith11]

Hi,
I have two eva4400 with continuos access.

I began to see link lost between the sites and the and the main site enter in the Replication Group transitioned to the Logging state because the alternate Storage System was not accessible.

Question:
Is possible that when the data replication log be full the EVA came down?.

How I test the latency between both sites?


Thank you.

W.S

[Víctor Cespón · Accepted solution]

If the log disk becomes full or the disk group where it is runs out of space, it will be discarded and the DRGroup marked for a full copy.
This means that once the DR link is restored, the full contents of the vdisks will be copied and not just the changes since the link was broken.
To monitor link performance, use evaperf drt -cont. Evaperf is in the same folder as Command View.

[Uwe Zessin]

The EVA _should not_ go down when the write history log (WHL) is full. It should stop logging and switch the Data Replication Group to a 'Full Copy'.

I have not tried it on an EVA-4400, but I have seen the (correct) behavior on a CA configuration with two EVA-3000.

[WilliamSmith11]

Then if the log become full all change since link was broken will be lost?

Whitch will be a good size setting for logging space?


Regards


W.S

[Uwe Zessin]

The contents on the WHL get lost, yes, but the data is written to the source virtual disk as well. A 'full copy' means that *all* data from the virtual disk will be copied again.

The WHL will provide a journal of all write activity. If the source fails, you have data on the destination that is not current, but at least crash-consistent (if the replication groups have been configured properly).

A fully copy sends all blocks again, but cannot protect against an error on the source.

[WilliamSmith11]

Thank you Uwe,




W.S