ESXi Host crashes due to connection problems to PetaSAN
d_17
1 Post
September 12, 2022, 9:44 amQuote from d_17 on September 12, 2022, 9:44 amWe have two Pools in our PetaSAN Cluster, one for our HDD Storage and one for our SSD Storage which are connected via iscsi disks to our vmWare Cluster. We are trying to get rid of our HDD Storage, which means we need to migrate all the disks from the HDD Pool to our SSD pool. This worked fine for smaller disks (50-400GB), but when we tried to migrate a 2.7TB disk, ESXi Hosts started having connection Problems to the Storage which made VMs unreachable and in conclusion the ESXi crash, with no other choice than to restart the host.
Does anyone have any idea what exactly could cause this problem and how to possibly prevent it?
Thanks in advance for any possible help.
We have two Pools in our PetaSAN Cluster, one for our HDD Storage and one for our SSD Storage which are connected via iscsi disks to our vmWare Cluster. We are trying to get rid of our HDD Storage, which means we need to migrate all the disks from the HDD Pool to our SSD pool. This worked fine for smaller disks (50-400GB), but when we tried to migrate a 2.7TB disk, ESXi Hosts started having connection Problems to the Storage which made VMs unreachable and in conclusion the ESXi crash, with no other choice than to restart the host.
Does anyone have any idea what exactly could cause this problem and how to possibly prevent it?
Thanks in advance for any possible help.
admin
2,930 Posts
September 12, 2022, 12:10 pmQuote from admin on September 12, 2022, 12:10 pmMost likely load on the HDD causing high latency delay which causes ESXi timeouts.
When moving vms, make sure the paths of both source and destination datastores are served by same PetaSAN nodes, see more detail in our VMWare guide under vmotion. This enables use of more efficient data movements, better performance and less load on the HDD.
Try to reduce other load during such transfers: client load if possible, turn off scrub and deep scrub in Maintenance UI or switch to slow, make sure you have no recovery healing in the background or switch its speed to slow.
Another approach is the use rbd cp command to copy the entire datastore to a new disk in the SSD pool, reducing load as above will also help.
You can also change the iSCSI lio_tunings under /opt/petasan/config/tuning to a less performance settings, see the templates directory.
Most likely load on the HDD causing high latency delay which causes ESXi timeouts.
When moving vms, make sure the paths of both source and destination datastores are served by same PetaSAN nodes, see more detail in our VMWare guide under vmotion. This enables use of more efficient data movements, better performance and less load on the HDD.
Try to reduce other load during such transfers: client load if possible, turn off scrub and deep scrub in Maintenance UI or switch to slow, make sure you have no recovery healing in the background or switch its speed to slow.
Another approach is the use rbd cp command to copy the entire datastore to a new disk in the SSD pool, reducing load as above will also help.
You can also change the iSCSI lio_tunings under /opt/petasan/config/tuning to a less performance settings, see the templates directory.
Last edited on September 12, 2022, 12:13 pm by admin · #2
ESXi Host crashes due to connection problems to PetaSAN
d_17
1 Post
Quote from d_17 on September 12, 2022, 9:44 amWe have two Pools in our PetaSAN Cluster, one for our HDD Storage and one for our SSD Storage which are connected via iscsi disks to our vmWare Cluster. We are trying to get rid of our HDD Storage, which means we need to migrate all the disks from the HDD Pool to our SSD pool. This worked fine for smaller disks (50-400GB), but when we tried to migrate a 2.7TB disk, ESXi Hosts started having connection Problems to the Storage which made VMs unreachable and in conclusion the ESXi crash, with no other choice than to restart the host.
Does anyone have any idea what exactly could cause this problem and how to possibly prevent it?
Thanks in advance for any possible help.
We have two Pools in our PetaSAN Cluster, one for our HDD Storage and one for our SSD Storage which are connected via iscsi disks to our vmWare Cluster. We are trying to get rid of our HDD Storage, which means we need to migrate all the disks from the HDD Pool to our SSD pool. This worked fine for smaller disks (50-400GB), but when we tried to migrate a 2.7TB disk, ESXi Hosts started having connection Problems to the Storage which made VMs unreachable and in conclusion the ESXi crash, with no other choice than to restart the host.
Does anyone have any idea what exactly could cause this problem and how to possibly prevent it?
Thanks in advance for any possible help.
admin
2,930 Posts
Quote from admin on September 12, 2022, 12:10 pmMost likely load on the HDD causing high latency delay which causes ESXi timeouts.
When moving vms, make sure the paths of both source and destination datastores are served by same PetaSAN nodes, see more detail in our VMWare guide under vmotion. This enables use of more efficient data movements, better performance and less load on the HDD.
Try to reduce other load during such transfers: client load if possible, turn off scrub and deep scrub in Maintenance UI or switch to slow, make sure you have no recovery healing in the background or switch its speed to slow.
Another approach is the use rbd cp command to copy the entire datastore to a new disk in the SSD pool, reducing load as above will also help.
You can also change the iSCSI lio_tunings under /opt/petasan/config/tuning to a less performance settings, see the templates directory.
Most likely load on the HDD causing high latency delay which causes ESXi timeouts.
When moving vms, make sure the paths of both source and destination datastores are served by same PetaSAN nodes, see more detail in our VMWare guide under vmotion. This enables use of more efficient data movements, better performance and less load on the HDD.
Try to reduce other load during such transfers: client load if possible, turn off scrub and deep scrub in Maintenance UI or switch to slow, make sure you have no recovery healing in the background or switch its speed to slow.
Another approach is the use rbd cp command to copy the entire datastore to a new disk in the SSD pool, reducing load as above will also help.
You can also change the iSCSI lio_tunings under /opt/petasan/config/tuning to a less performance settings, see the templates directory.