VMware Cloud Community
marinador
Enthusiast
Enthusiast

Continuous consolidations in server

For several days the same servers (mainly one of them) have been giving constant consolidations after making the corresponding copies with Veeam Backup & Replication 8.

After several attempts following the documented recommendations of VMWare the one that has resulted was to add the parameter snapshot.asyncConsolidate.forceSync to TRUE in the affected machines and the problem was solved temporarily. Today has consolidated the consolidation of this server (a SQLServer of about 400GB) and does not finish consolidating it or removing the Veeam snapshot and has remained in a kind of infinite loop.

Additionally I have noticed a brutal increase in the latency in the Datastores after the beginning of these episodes of daily consolidations. I have reviewed the logs of the ESX and only observe what I leave in the capture.

Help please, it is a production server and it is almost not working all day consolidating

Reply
0 Kudos
5 Replies
daphnissov
Immortal
Immortal

Need more information from you here. What version of vSphere (vCenter and ESXi, please)? Is this/these VMs running on local storage to ESXi? If so, what is it and its characteristics? Generally speaking when I've had and seen these issues before it's due to poorly performing storage. When Veeam (or any other vendor who uses the VADP for its backup methodology), calls to remove a snapshot they all use the same method and from there on it's up to the infrastructure to complete that operation. If that operation is taking a long time or failing it's usually because of the underlying storage. And in looking at your first screenshot, it appears (inferring from the name) the datastore is a RAID-5 volume using NL-SAS drives. Depending on how many drives you have in use, this is likely the problem due to the write penalty incurred by RAID-5/6. The performance is terrible on write operations especially on smaller blocks. The only fix in these cases is to improve storage performance by moving to more capable hardware.

Reply
0 Kudos
marinador
Enthusiast
Enthusiast

VCenter 5.5 and the VMs run on SASNL drives...

Reply
0 Kudos
daphnissov
Immortal
Immortal

Then your problem is likely due to poor storage performance. Being on that version of vSphere also means you don't have better snapshot performance that came with vSphere 6.0, and with end of support looming for vSphere 5.5 in just a couple weeks you really need to get off that version like right now.

Reply
0 Kudos
marinador
Enthusiast
Enthusiast

We are aware that we have to update with these times, two weeks ago, previously the copy systems did not generate as many consolidations ...

Reply
0 Kudos
daphnissov
Immortal
Immortal

Storage conditions change and it's hard to account for that unless your storage is well performing which it sounds like it isn't.

Reply
0 Kudos