Running a cluster on NFS
One of the guests (Ubuntu 20) does also have shares on 2 nfs servers that are also mounted on PVE for f.i. iso or images. Both PVE and guest are on the same network 10.0.1.x and are using the same ip's (10.0.1.8 & 10.0.1.50)
After a network failure a couple of weeks ago (switches accidentally switched off) this guest suddenly has massive problems to keep the nfs connections acceptable. lots of timeouts, up to minutes, for a simple ls or cd.
When trying a ping to both the nfs servers from the guest there's substantial packet loss. However, when i ping from the PVE shell where the guest resides to the same ip's there's zero packet loss. Also no problems from other machines on the network.
So it looks like there are network problems. But only from that guest, not from PVE itself. We already tried swapping switches but that didn't make any difference.
All cluster guests are running fine while having their disk on the shared storage. So i suspect there's an issue with that guest in conjunction with PVE.
Migrating the guest to another PVE in the cluster makes no difference.
Anyone any idea what is going on here ? Cluster is running 7.3 and 6.4, tried both, no difference.
One of the guests (Ubuntu 20) does also have shares on 2 nfs servers that are also mounted on PVE for f.i. iso or images. Both PVE and guest are on the same network 10.0.1.x and are using the same ip's (10.0.1.8 & 10.0.1.50)
After a network failure a couple of weeks ago (switches accidentally switched off) this guest suddenly has massive problems to keep the nfs connections acceptable. lots of timeouts, up to minutes, for a simple ls or cd.
When trying a ping to both the nfs servers from the guest there's substantial packet loss. However, when i ping from the PVE shell where the guest resides to the same ip's there's zero packet loss. Also no problems from other machines on the network.
So it looks like there are network problems. But only from that guest, not from PVE itself. We already tried swapping switches but that didn't make any difference.
All cluster guests are running fine while having their disk on the shared storage. So i suspect there's an issue with that guest in conjunction with PVE.
Migrating the guest to another PVE in the cluster makes no difference.
Anyone any idea what is going on here ? Cluster is running 7.3 and 6.4, tried both, no difference.