Search results

  1. T

    Why we can't have more than 4094 vlan tag ID on OVS with GRE overly?

    That's correct, but doesn't OpenVSwitch supports IEEE_802.1ad ?
  2. T

    Why we can't have more than 4094 vlan tag ID on OVS with GRE overly?

    I setup OVS switch on 10 nodes which are interconnected via GRE tunnels, but I can see Proxmox is not letting us create more than 4094 vlans because tag ID larger than 4094 is not validated in Proxmox interface or API. Is there any restriction based on GRE overlay that prevent us having more vlans?
  3. T

    Prometheus guest exporter

    As far as I know recent versions of GuestAgent support executing any commands, therefor it should be possible to monitor guests from host, is that true?
  4. T

    ovs rstp_designated_path_cost for GRE

    In proxmox documents I can see it's advised to set rstp_designated_path_cost for physical ports, but as I want to create a mesh network with openvswitch containing 20 nodes which all are connected together via GRE is it necessary to set path_cost for GRE too? I've tried command below but it...
  5. T

    Prometheus guest exporter

    I wonder if there's any exporter that gather guest usage information from inside of them via GuestAgent and export them to prometheus? If not is it reasonable to use GuestAgent for gathering resource usage of VMs?
  6. T

    iptables is not picking tap interface packets

    It fixed by using "-m physicaldev --physdev-in tap101i0" instead of -i, and it's because tap is layer 2 not layer 3
  7. T

    iptables is not picking tap interface packets

    I've used a simple iptables rule to test some idea, but I can see it's not picking any packet on tap interface. Is that normal? root@node01:~# iptables -I FORWARD -i tap101i0 results: root@node01:~# iptables -L FORWARD -v -n Chain FORWARD (policy ACCEPT 0 packets, 0 bytes) pkts bytes target...
  8. T

    OVS on many servers

    So as I've L2 connectivity between all nodes, how can I avoid establishing GRE between all pairs of servers?
  9. T

    OVS on many servers

    My servers are all connected via layer2 switches, but I can't add so many vlans on them.
  10. T

    OVS on many servers

    I've a cluster of 40 Proxmox nodes and I want to start offer private network to my clients, Each client of me may have few VMs on different nodes, and it's not static, they may add/remove VMs. My question is if I want to setup OVS based private networking it's necessary to add GRE for each two...
  11. T

    Create RBD image based on another image

    I've build a cloned VM using linked clone from Proxmox interface, but consider when I want to rebuild that VM, I will need to remove the primary disk of that VM and add an other one which is based on an other base image, ( for example changing the OS of that VM from Ubuntu to CentOS ) Is there...
  12. T

    OpenSM is not detecting ib ports

    And this also got fixed by "modprobe xprtrdma" I've add them to /etc/modules to auto load them after reboot
  13. T

    OpenSM is not detecting ib ports

    After running "modprobe ib_umad" on both Proxmox nodes, OpenSM detects the ports but still the link is in initialization status: service opensm status ● opensm.service - LSB: Start opensm subnet manager. Loaded: loaded (/etc/init.d/opensm; generated) Active: active (exited) since Fri...
  14. T

    OpenSM is not detecting ib ports

    I'm connecting two Proxmox servers with ConnectX-3 cards, they can see each other in Ethernet mode but in ib mode, the link remains in Initiating state despite physical link up status. Also in opensm logs I can see that it cant detect ib ports while there are active. and both cards are in the...
  15. T

    Poor randwrite fio performance result

    Thanks, you are right indeed. I've figured it out by "--image-shared" option when creating fio image, like code below: rbd -p rbd create --size 2048 fio_test --image-shared
  16. T

    Poor randwrite fio performance result

    I've a SSD based Ceph cluster with 3 nodes, the read IOPS is about 250K with 96 parallel fio jobs ( running from 3 different nodes ), the reasults are fine. But the Write performance is not more than 2k IOPS when more than 1 parallel jobs is running. ( with only one job I can reach 15K write...
  17. T

    [SOLVED] Mellanox ConnectX-3 is not detected by Proxmox

    The card was faulty, Our vendor has changed that card for us. Sorry to bother.
  18. T

    [SOLVED] Mellanox ConnectX-3 is not detected by Proxmox

    Hello, I'm going to use Mellanox ConnectX-3 for Proxmox 5.4 but the kernel is not detecting it, while it's being detected fine by Windows on the same server. Any help on how to make it work? lspci | grep Mellanox returns empty result pveversion -v proxmox-ve: 5.4-1 (running kernel...
  19. T

    Monitor in Unknown status

    I guess it's monitor connection issue, because monitor service is up and running. Any help on how to troubleshoot it would be much appreciated.
  20. T

    Monitor in Unknown status

    I had a two node ceph cluster, It was working fine, but after adding new node to ceph cluster and there for joining it to proxmox cluster, one of my previous monitors went into unknown status despite it's running. Also the new monitor is in unknow status too. Please check the attached picture...