Newbie question on HA

[robnicholson]

I've got my lab working with a 2-node HA and did the following:

1. Started a 3GB file copying from a test file server using an iSCSI connection to both nodes
2. Stopped the StarWind service on the second node
3. The file copy completed - yeah!

I've now restarted the service on the second node so obviously, it's out of sync.

The question: does the node have to completely resynchronise the disk or does it know which blocks are out of sync and will only re-sync that bit?

The question is really how long it takes for a node that was off-line for a short amount of time to come back up?

Cheers, Rob.

[anton (staff)]

The answer is - it depends. We keep a bitmap of a changed blocks so with some amount (small) changed we do a fast sync doing only changed block wire. However if some threshold value approached it makes sense to do a full sync as it would be faster - streaming huge amounts of data in a pipeline mode is faster then doing read-send-write-wait sequence typical for a changed block tracker syncs.

[robnicholson]

Thanks. Any idea what that threshold is? Cheers, Rob.

[Anatoly (staff)]

99.9% StarWind records everything , so it will synchronize only changed blocks. The clean full sync will appear only if 100% of data on HA mirrors will be different.
Also, I`d like to share the info below with you:

The Full synchronization is starting instead of Fast in next cases:
1)If HA device was configured to use write-back cache and one of the server was turned off not correctly (i.e. hard reset, power outage, etc);
2)If the errors were detected during the writes on the disk;
3)If partner, that should be the source of synchronization has the state another from "Synchronized";
4)If initial synchronization was interrupted due to any reason;
5)If extension of the HA device was performed (in this case the synchronization will be performed only after service restart)
6)If the Partner that should be the source of synchronization gained "Synchronized" state after initiating "Mark as synchronized" command
7)If StarWind service was turned off on all servers

Synchronization will not start automatically if:
If the errors were detected during the writes on the disk;

I hope that makes sense.

[robnicholson]

Thanks for the reply. Can I ask a few more questions here:

>5)If extension of the HA device was performed (in this case the synchronization will be performed only after service restart)

Is that EVER performed or performed whilst one of the nodes was offline? I'm hoping it's the later as expanding the size of a device will be a pretty common occurrence IMO if you are using flat images.

>7)If StarWind service was turned off on all servers

Hmm, what happens if there is planned outage here and you do this:

a. Shutdown server node #2
b. Shutdown server node #1
c. Do some work (e.g. add more RAM)
d. Power-up node #1
e. Power-up node #2

Are you saying that it'll do a full re-sync here?

Cheers, Rob.

[Anatoly (staff)]

5. The extension will not be performed - that option is blocked due to data security reasons if one of hte HA partners is offline.
7. The options when full sync will appear applies to *.img based HA devices only. The LSFS devices will run only fast synchronization. I hope that makes sense.