Search results

i have read that before and read it again i'm not sure which one i have to use Adaptive transmit load balancing (balance-tlb): Linux bonding driver mode that does not require any special network-switch support. The outgoing network packet traffic is distributed according to the current load...

Thanks very nuch for the information for example that i can created the private network for the Ceph by using node1# pveceph init --network 10.10.10.0/24 How i can bond the other link?

Hello, Just have some questions regard Ceph. 1 - 10Gb switch Layer 3 or 10Gb switch Layer 2 will works? 2 - I have Dual 10 Gbit ports possible to use both private Ceph cluster? for better performance? Thanks

there is another kworker % CPU and I/O issue as well but not shows for the moment it come every few days and gone after reboot and back. root@lxx:~# cat /proc/interrupts CPU0 CPU1 CPU2 CPU3 CPU4 CPU5 CPU6 CPU7 0: 15 0...

I revers my reply it seems pve-kernel-4.10.17-5-pve kernel have issue as well TID PRIO USER DISK READ DISK WRITE SWAPIN IO> COMMAND 3829 be/4 root 3.56 K/s 0.00 B/s 0.00 % 99.99 % [kworker/u16:4] 6629 be/0 root 3.93 M/s 118.71 M/s 0.00 % 91.26 % [loop25]...

I can confirm that ksoftirqd fixed in 4.10.17-5-pve i tested in one node.

I going to install it now but do this have kworker issue?

I had the same issue as well also sometimes issue with kworker eat my CPU and I/O as well

Do you have LXC and KVM on that node? or KVM only?

I still have this pain in my work even it sometime come with PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 28627 root 20 0 0 0 0 R 99.7 0.0 3652:46 [kworker/u16:0] This kworker + ksoftirqd eat my i/o and CPU each upgrade the kernel...

How about PCIE SM961 as Journal? It seems enterprise!

Thanks very much for letting me know i will do it as 3 however i still have issue to know what kind of Journal need to use for my case.

i will start with 3 node and looking for something 100VM or 150VM maybe more depend on how much load i do get. i looking to do about 70TB of space . for what i understand replicated 2 will be fine with me .

Any help would be appreciate it if possible recommended such PCIE for Journal i can go with 2 x Journal one PCIE and one SATA drive. i did some google but got nowhere i checked https://www.sebastien-han.fr/blog/2014/10/10/ceph-how-to-test-if-your-ssd-is-suitable-as-a-journal-device/ however...

So from what i understand that the journal will works as SSD cache for my SATA OSD so i need fasted i can get and with more enterprise version? The SM825 just SATA is there is better as PCI instead? as Journal and what is better using LSI 9210-8i HBA to connect all 8 bays or just using...

So 1 x Samsung 960 PRO Series - 512GB PCIe NVMe would be better than 2 of those? i also see SanDisk FS1-004-320-CS-0001 Fusion IO Drive 320GB ?

KVM Storage my reason to use the SATA enterprise Drives that i need huge of amount of space which SSD can't offer for cheap price.

So the MyDigitalSSD BPX bad choice? Samsung SSD 950 PRO cost around $200 or more per piece

Hello, Planing to setup small ceph but i do care much for i/o speed my setup will be following 3 x U2 72GB RAM SSD SM863 for Proxmox installation 3 or 4 x 4tb enterprise drives. planned to add up to 5 of them later 2 x 120GB MyDigitalSSD BPX 80mm (2280) M.2 PCI as journal Dual 10 Gbit Intel...

i finally fix that problem with last update however something shows now which is really very annoying PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 23254 root 20 0 0 0 0 R 100.0 0.0 2151:08 [kworker/u16:3] doing high i/o and eat the CPU...