This post is more than 5 years old
5 Posts
0
32683
Setting Up SyncRep
We've decided to change our setup a bit. Previously we had 2 identical virtual hosting rigs in 2 different server rooms on campus. Both consisted of a PS4100 and 3 Dell servers in a failover cluster with a CSV stored on each SAN. Our main rig replicated VM's to the backup rig via Hyper-V replication. Because our needs have expanded, the resources on 3 host servers can no longer hold the load and have enough to spare for failover. So we have moved the 3 backup hosts to the main server room and will be adding them into the existing cluster, then replicating the SANs, hopefully using SyncRep. I have added the backup SAN into the group with the main SAN and created a separate storage pool, but when I try to enable SyncRep, we discovered our problem.
The volume we have set up for the CSV is 6.5 TB and we only have enough extra space to set up a 6% snapshot reserve. I figured this would be ok since the minimum requirement for SyncRep is 5%, but when initial synchronization kicks off, it fills up the snapshot reserve and then throws an error saying the snapshot reserve is exhausted and starts the process over again. I'm not sure where to go from here, is it possible to do a manual initial synchronization, similar to traditional replication with the manual transfer utility?
Here is a snip of what I'm seeing:
Thanks in advance for any advice you might have for us.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
November 6th, 2014 07:00
Hello,
There is no manual sync rep. I would start by thin provisioning the volume to provide more space for snapshot reserve. I would also try to do that for other volumes to provide additional free space as well.
You have the initial 4+TB to sync, plus new writes coming in so additional snapshot reserve is needed.
What's the interconnect between the source and alternate pool? Are they on the same switches?
If that still fails I would suggest opening a support case.
Regards,
jmsmith76
5 Posts
0
November 6th, 2014 11:00
I talked to support and it is necessary to have 100% (preferably more) snapshot reserve, at least for the initial sync. We were able to clear off a couple TB of old servers to free up some space. Then we enabled thin provisioning to allow more space to be allotted for snapshots. I will be kicking off the initial sync this evening after network/server traffic slows down for the day and will post back once we get it sync'd up successfully.
jmsmith76
5 Posts
0
November 11th, 2014 08:00
I seem to have accidentally deleted the post with the info I got from support, so here it is again:
I talked to support and it is necessary to have 100% (preferably more) snapshot reserve, at least for the initial sync. We were able to clear off a couple TB of old servers to free up some space. Then we enabled thin provisioning to allow more space to be allotted for snapshots. I will be kicking off the initial sync this evening after network/server traffic slows down for the day and will post back once we get it sync'd up successfully.
jmsmith76
5 Posts
0
November 6th, 2014 08:00
We've got redundant PowerConnect 6224 switches on each end, with dedicated 1G fiber link between them. Any idea how much snapshot reserve we'll need for 4+ TB of data? Do we have to have 100% reserve?
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
November 6th, 2014 19:00
Thanks for updating us. It's also dependent on the latency on that link. That will impact the sync and normal I/O writes once it's finished the initial sync.
Regards,
jmsmith76
5 Posts
0
November 11th, 2014 08:00
I wanted to give it a few days before posting an update. So far, we are running just fine. The first morning after syncing, we had a few users complain about Outlook running slow on their computer (Exchange is hosted on the cluster), but I haven't seen or heard of any issues since then. I'm thinking that the sync wasn't quite finished at that time. Other than that, no performance issues that I've noticed at all. It did take probably 12+ hours for the intitial sync to happen, which was around 2.5 TB's of data. Thanks for the help Donald. Maybe this thread will help someone else out in the future.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
November 11th, 2014 08:00
That's great news!! I'm glad I could help a little bit anyways.
Regards,