gem.cafe

new guy just joined, need help & suggestion

Recommended Posts

hallo,

 

vmware also has logs so you might have a look there and why it was not accesible anymore

 

https://docs.vmware.com/en/VMware-vSphere/6.5/com.vmware.vsphere.storage.doc/GUID-0114693D-94BF-4D0E-9BA4-416D4A51A5A1.html

 

"The RDM is not available for direct-attached block devices or certain RAID devices. The RDM uses a SCSI serial number to identify the mapped device. Because block devices and some direct-attach RAID devices do not export serial numbers, they cannot be used with RDMs."

 

not shure how that comes into view when using raid0 devices (for every single disk) from the hp controller

you might consider to use ready made raid devices (like raid5) from the controller and have just one rdm and using it as basic volume in dsm or (if it worsks as advertised) use the hba mode, then i'd expect the disk should not be present as hp "logical" (as seen in the picture) it should be seen with all infomation like disk vendor, at least in vmware?

 

my suggestion would be hba mode and rdm (that way you could use the disks on any other controller that work as hba) or use a hp (hardware raid) over all disks (like raid5), create a vmfs datastore and use a vmdk file for a basic volume in dsm, the second option is kind of default use of vmware and if you leave same space on the volume (vmdks can be extended later) you can use it for other vm's like linux/windows vm's

i would try to not use hp single disk raid0 volumes in controller raid mode as disks, hba+rdm and let dsm do raid or hardware raidX and let the controller do the redundancy thing and let dsm use it as basic volume

 

Share this post


Link to post
Share on other sites

thanks for your suggestion, IG-88, I'm planning to do something like your suggestion

 

my purpose for the last round is just for testing & want to see what dsm does about the drives actually 

I killed the whole setup already, and will do one more round of trial tonite as following...

 

set the b120i as ahci, with 2 x 60gb ssd (all basic, for vm data store)

set the p420 as hba, with 2 x 120gb ssd + 2 x 1tb hdd (all basic, will try to passthru the whole controller to dsm from vm)

 

* a guy from web who is also using gen8+dsm, he said, esxi's hp image has problem (vm hang) if rdm is using, but vmware image is fine

 

if everything alrite this time, and able to run a week, then will go ahead to buy 4 x 4tb (or bigger) drives for data moving

Share this post


Link to post
Share on other sites
On 11/11/2017 at 2:05 PM, gem.cafe said:

set the p420 as hba, with 2 x 120gb ssd + 2 x 1tb hdd (all basic, will try to passthru the whole controller to dsm from vm)

 

thats not giong to work as dsm does not have a working driver for the p420 (means also no baremetal install)

 

On 11/11/2017 at 2:05 PM, gem.cafe said:

* a guy from web who is also using gen8+dsm, he said, esxi's hp image has problem (vm hang) if rdm is using, but vmware image is fine

 

as i'm not using this hardware and even use baremetal for my own system, i guess you should consider the advice

 

On 11/11/2017 at 2:05 PM, gem.cafe said:

4 x 4tb (or bigger) drives for data moving

 

bigger is better as the port count and space in the case is limited, but i also look at the price per TB and disc statistics of the company backblaze (https://www.backblaze.com/blog/), reliability is a factor (in sum up it takes a lot of time when a disk failes when you take all the hassle into account including phone calls, shipping, ...)

3 years ago i ended up with 4TB discs, i'd expect it to be bigger discs if evaluated now

 

Share this post


Link to post
Share on other sites

:shock: can't bare metal bcoz no driver for p420, can't pass thru bcoz no driver for p420, can't rdm bcoz get hang easier... then... really need to give it up!?

 

anyway, I will give it a try!

just re-installed the esxi; modified the img's cfg file (s/n, mac...) + added extra.lzma 3.2.

 

yes, I hope to grab 4 x 8tb ;) getting cheaper now

toshiba 8tb MG05ACA800E around usd$200 today, rumours said, the price going down bcoz MG06 10tb will hit the market soon

 

** windows 10 on-to-go is very good for mac user who needs to use windows' tools, osfmount, notepad++

 

Share this post


Link to post
Share on other sites
2 hours ago, gem.cafe said:

 can't bare metal bcoz no driver for p420, can't pass thru bcoz no driver for p420, can't rdm bcoz get hang easier... then... really need to give it up!?

no, you use p420 hardware raid with the esxi driver, format the relsuting raid volume in esxi as vmfs (aka datastore in vmware) and place a vmdk file (virtual disk) in that volume, the vm for dsm uses a scsi/sas controller (lsi or for better performace paravirtual) and dsm sees one big disk that will be used as basic (no raid) inside dsm, redundency is done by the hardware raid and the esxi driver, esxi has to inform you about a disk not working

 

2 hours ago, gem.cafe said:

+ added extra.lzma 3.2.

no need as vmware is supported by the default extra.lzma

 

 

 

2 hours ago, gem.cafe said:

** windows 10 on-to-go is very good for mac user who needs to use windows' tools, osfmount, notepad++

there are also native ways dealing with that (and finding a unix compatible editor on a bsd "clone" like mac os x ....)

http://www.gingerbeardman.com/dd-gui/

https://itstillworks.com/mount-img-mac-8453471.html

Edited by IG-88

Share this post


Link to post
Share on other sites

I really have bad luck every moments...

 

2 x 1 tb hdd over-heated in a small case, facing vmware known issue (memory limit doesn't accept the negative value, -1048576), vm able to see the drives from hba but couldn't make datastore from them... shxt!

 

negative.jpg.0a08fda945dd79ac3d931a6ad394a6a3.jpg

 

19 hours ago, IG-88 said:

no need as vmware is supported by the default extra.lzma

I didn't know that, put it there just in case

 

19 hours ago, IG-88 said:

no, you use p420 hardware raid with the esxi driver

hahaha... u got thousand-miles eye? I need to switch back the card to raid mode, but I will try set all drives to raid-0

 

19 hours ago, IG-88 said:

there are also native ways dealing with that

no, I tried many times by using my mac (double click, disk utility..) to open the img, failed everytime, that's why I switched to windows and modified the img file

 

img.jpg.d3d506d87f1b4aa23a505530f21b20c4.jpg

Share this post


Link to post
Share on other sites
5 hours ago, gem.cafe said:

hahaha... u got thousand-miles eye? I need to switch back the card to raid mode, but I will try set all drives to raid-0

will be difficult to keep track of all this volumes and sizes, you will have to make shure the vmdk files have all the same size than

and to keep it predictable (like if i pull this disk ths virtial disk is the one to fall aout of the raid in dsm) you should make good names for datastores and dokumentation what physikal disk is what data store

hba mode would be better, maybe the driver in esxi is to old, vmware also has hp custom iso's for esxi and if you can't get them there are also hp drivers packages that can be installed (hp olso has a software repository that can be used in vmware, but not shure if that is possible without comecial license and vcenter server)

 

PS: dont forget to activate the write cache of the disks, the HP default might be disabled write cache as it is safer in case of power failsures

 

Share this post


Link to post
Share on other sites

after I read more & more articles from web, not sure I got understand or not, here is what I learned...

 

passthru = rdm, vmware esxi only has rdm (from vm kb), but it gets 2 ways, virtual & physical

 

virtual: vmkfstools -r / physical: vmkfstools -z

 

both of them need to format the target drive to vmfs, then use the either command/parameter on above to make a vmdk file

just not sure the rdm creation from esxi's webgui is virtual or physical, need time to find out :D

 

18 hours ago, IG-88 said:

you should make good names for datastores and dokumentation what physikal disk is what data store

18 hours ago, IG-88 said:

PS: dont forget to activate the write cache of the disks

I did long time ago ;)

 

Share this post


Link to post
Share on other sites
4 hours ago, gem.cafe said:

passthru

 

in the context of the controller that would be using VT-d to passthru the controller as pci-e device to the vm so that dsm vm sees the controller, loads the driver and has direct disk access

 

you cant realy passthru a disk, you can do rdm it and thats as close as you can get, no s.m.a.r.t. data or temp from disk, just a sector mapping to the real disk

when using vmdk files as (virtual) disks  you can male snapshots, then a new file is created an all new wriiten data is in that file, if you have a rdm then there is no space to write snapshot data, there cames the virtual rdm into play, you can make a snapshot but additional to the rdm device you will need a vmfs datastore to write the snapshot data

5 hours ago, gem.cafe said:

both of them need to format the target drive to vmfs

no, the rdm device is directly (sector wise) written by the vm and can have any format, vmware does simply take sectors (lba) and writes them to disk, vmfs will only be needed when a virtual rdm is used (for storing the snapshot data -> see above)

5 hours ago, gem.cafe said:

just not sure the rdm creation from esxi's webgui is virtual or physical, need time to find out

in the gui i have seen you could choose when creating rdm, when nothing is seen then it will be just a normal rdm (not virtual), if creating a virtual rdm you will have to make a choise where to write the snapshot data

 

Share this post


Link to post
Share on other sites

thank you so much for your help, IG-88, you helped me a lot to complete this task! :19_kissing:

 

setting for the last round of trial…

ms gen8, p420 raid mode, 2165L v2, 16gb

jun’s loader v1.02b, image ds3615xs, pat dsm v6.1.4-15217-1

 

after created few vmdk files for the drives by using cli (vmkfstools -z), the dsm is able to see all drives finally

 

now, the dsm is up & running, hard drives’ arrangement is almost the same as I planned, 1 ssd drive for vm datastore, 2 ssd drives for cache, 2 hdd drives for temporary storage (bt), 4 extra new data drives will be coming soon

 

unfortunately,… no drives’ smart data can pass to the dsm, it makes me a bit uncomfortable :7_sweat_smile:

 

 

** tried to load the image ds3715xs, but failed to locate the diskstation by using find.synology.com, as ds3615xs supports 2-core cpu only, so will do re-set the vm with 2-core cpu only

 

01.jpg.0940704673e27db7244ccd65ebbb2ca2.jpg

02.jpg.c14435fa4e58d44b1d2bc611c899ede2.jpg

03.jpg.0c8e5df849ff879054b38ba3fe680f82.jpg

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now