How do I make replicated volumes writable?

[elangley]

Hello All,

I use FreeNAS replication to backup vSphere virtual machines to my datacenter.

Just upgraded to 9.3 Stable from 9.2.1

In 9.2.1 I had replicated snapshots mounted and shared with my vSphere server. I used this to run quick tests to verify that replicated virtual servers would startup and then shut them down. Starting a virtual machine requires a writable volume, even if it gets overwritten later, which is actually what I want to happen.

Now, with replicated volumes set to read only, I have to clone the volume, make a share, mount that share in vmware, add the machines I want to test, blah blah blah. Much wasted time.

How do I make replicated volumes writable?

~eric

 

[depasseg]

There isn't a way within the FreeNAS UI. In fact there were issues having the replication target read-writable, which is, I believe, why they changed to to be read-only. You can try using the CLI to unmount and mount re-write, but FreeNAS might change it back (possibly after the next replication or reboot). Please be aware, that this isn't tested, nor advised and may require you to destroy your replication target and start replication from scratch.
https://docs.oracle.com/cd/E23824_01/html/821-1448/gaynd.html

 

[elangley]

Thanks depasseg,

According to the Oracle docs the method is;

Shell: zfs set readonly=off tank/dataset

In my environment I can confirm that a replicated dataset is writable after changing this setting. I will test again after a currently running replication finishes to see if the readonly=off setting holds.

I appreciate your concern about writing to the replicated volumes. While I am not sure about how 9.3 will work out with 9.2.1 I never had an issue with replication failing after briefly mounting the volumes with writes. Again, more testing will tell.

Thanks again,

~eric

 

[depasseg]

There have been a LOT of changes with snapshots and replication over the past 4-5 months on 9.3.1. It's been quite a ride.

Good luck!

 

[elangley]

The release note for 9.3 state the following about replication:

"* The replication code has been substantially rewritten and can now
recover from out-of-sync replications it far better fashion."

What are "out-of-sync replications"?

 

[depasseg]

I'm not sure, but I do know that every once in a while suffering would happen that would cause me to destroy the target and start replication from scratch. Things have been better the past month or so.

 

[Eduardo Gaibor]

hi elangley, what was the results of your tests?
currently i am having the same problems that you mention

 

[elangley]

I was able to get it working. No issues so far as we never make changes to the destination. For actual use taking a snapshot and mounting it is still the best method.