[PVE-User] openiscsi with Proxmox
Fábio Rabelo
fabio at fabiorabelo.wiki.br
Thu Jan 2 19:10:28 CET 2014
I am not an expert on ZFS, the best place to ask anything is the
OmniOS forum ...
But that something I can say, RAM memory !
The people who give me help when I was implementing says this :
16GB of RAM for boot, PLUS 1 GB of RAM for each 1TB os raw disk space !
And one more thing, the Sata ports in your system are 3 GB/s or 6 GB/s ?
It makes a lot of difference,
mainly in the SSD ZIL/L2ARK disks ...
Fábio Rabelo
2014/1/2 Muhammad Yousuf Khan <sirtcp at gmail.com>:
>
>>
>> As an example, in the test fase, I removed one disk from the cage with
>> the system running !!
>>
>> The hot spare previously configured assumed the task, and after I
>> reconnected the disk, I was able to make it the new hot spare, all in
>> the web UI .
>>
>
> actually i am not worried about the drives that is quite easy even we change
> it via ZFS command line and work great. however what i am afraid of is the
> OS. if some thing happens, like system crash etc. Lets say if Ethernet fails
> due to an hardware issue. that would be a problem. but yes if i use 2
> napp-it boxes and send replicated snapshot that i can save my save from the
> issue.
> but
>
> i have a Dell Workstation, with 12GB ECC RAM and Xeon 2x3.0 Quad core. with
> 4 interface Ethernet LAN (Intel). 2 SSDs for ZIL and 6 Sata drives with
> different sizes to test the RAID.
>
> do you think it is a good system to use in production?
>
>
> i can not reach 4x1GB thoughput from intel ethernet. i at least need 1 GB
> throughput from each Ethernet interface.
> what do you think why it would have been happening?
>
> Since how long you have been using this in productions?
> which hardware specs you are using for production?
> are you using the dual USB boot nappit or you are booting it from internal
> drive?
>
> thanks
>
>> Fábio Rabelo
>
>
More information about the pve-user
mailing list