storageman
10 Posts
Quote from storageman on July 30, 2021, 4:52 am
We have identified that our 3rd node ethernet 10gb 2-port stopped working. Both LEDs are not illuminated.
We are thinking for replace with new adapter.
What is best way to do & It is possible without downtime.
Thanks,
We have identified that our 3rd node ethernet 10gb 2-port stopped working. Both LEDs are not illuminated.
We are thinking for replace with new adapter.
What is best way to do & It is possible without downtime.
Thanks,
admin
2,930 Posts
Quote from admin on July 30, 2021, 12:29 pm
1 node down will not cause the cluster to be down if other nodes are ok. make sure you check the new interfaces get named correctly from the blue node console menu you can check / rename interface names.
1 node down will not cause the cluster to be down if other nodes are ok. make sure you check the new interfaces get named correctly from the blue node console menu you can check / rename interface names.
seanp
9 Posts
Quote from seanp on August 10, 2021, 4:33 pm
It is difficult to discern whether you mean the two port adapter is faulty, or port 2 itself is faulty (I have had this happen on this exact NIC). If its just port two, then cable into port 1 and switch it in the web interface. If it is the card, you'll obviously have to power down the host to swap it out.
It is difficult to discern whether you mean the two port adapter is faulty, or port 2 itself is faulty (I have had this happen on this exact NIC). If its just port two, then cable into port 1 and switch it in the web interface. If it is the card, you'll obviously have to power down the host to swap it out.
storageman
10 Posts
Quote from storageman on August 21, 2021, 5:48 am
Quote from admin on July 30, 2021, 12:29 pm
1 node down will not cause the cluster to be down if other nodes are ok. make sure you check the new interfaces get named correctly from the blue node console menu you can check / rename interface names.
I have replaced the nic card but in CEPH health long heartbeat ping warning is showing.
Also,
As you told one node does not effect cluster but I our environment if we shutdown one node the some pool/ISCSI disk are invisible. please guide
Thanks
Quote from admin on July 30, 2021, 12:29 pm
1 node down will not cause the cluster to be down if other nodes are ok. make sure you check the new interfaces get named correctly from the blue node console menu you can check / rename interface names.
I have replaced the nic card but in CEPH health long heartbeat ping warning is showing.
Also,
As you told one node does not effect cluster but I our environment if we shutdown one node the some pool/ISCSI disk are invisible. please guide
Thanks
storageman
10 Posts
Quote from storageman on September 17, 2021, 1:20 pm
Team,
Any updates?
In our production if one node goes down or one node shut down for any maintenance task. Some pools/iSCSI disk stopped working. We have set to 2 replicas for pools. how are can set proper cluster or HA PetaSAN cluster.
Team,
Any updates?
In our production if one node goes down or one node shut down for any maintenance task. Some pools/iSCSI disk stopped working. We have set to 2 replicas for pools. how are can set proper cluster or HA PetaSAN cluster.
admin
2,930 Posts
Quote from admin on September 17, 2021, 1:30 pm
what are the pool size and min_size values ?
what are the pool size and min_size values ?
admin
2,930 Posts
Quote from admin on September 18, 2021, 6:47 am
To have the pool highly available, you should increase your size to 3 while min_size remains at 2
To do this make sure your OSDs have the capacity for this extra replica so they should be currently at around 50% else you need to add more disks. Depending on hardware + current load + how much data is stored, you may need to lower/adjust the recovery speed from maintenance tab so to not put too much load on your system
To have the pool highly available, you should increase your size to 3 while min_size remains at 2
To do this make sure your OSDs have the capacity for this extra replica so they should be currently at around 50% else you need to add more disks. Depending on hardware + current load + how much data is stored, you may need to lower/adjust the recovery speed from maintenance tab so to not put too much load on your system
