Jump to content

unRAID Server Release 5.0-rc16c Possible Spindown issues under ESX and possible enhancements regardi

TheDragon

By TheDragon
in User Customizations

Recommended Posts

madburg Collaborator

madburg

Posted
Posted

Dude u are going about it the wrong way, the mass users are not linux people and they don't understand anything you're saying, even if you have very valid points.

Most people were happy ur account was deleted, and don't understand the lost that comes with it. The few that understand you, still don't agree with the shove it down the mouth approach.  So as someone who appreciates what u bring to the table, I can only said find another approach or its only going to get worse  :'(

Link to comment
  • Replies 64
  • Created
  • Last Reply
WeeboTech Veteran

WeeboTech

Posted
Posted

This is great information to have, especially as free support techs. LOL.

 

- if array not started we don't check spin-down.  This is by design.

- checking temps, spinning up/down never done with disabled disks, again by design (because could cause flood of messages to syslog and severe slow down due to error recovery because, well, the drive is bad).

 

In all seriousness, can you add a message to the main display whenever automated spin down timers are disabled.

Perhaps something at the top bar in RED or CAPS.

 

It helps to know so that people know it's safe to run long running drive maintenance tasks without being interrupted.

I'm thinking about smart -t long tests which can run for over 5 hours or so.

 

Link to comment
madburg Collaborator

madburg

Posted
Posted

Just in case anyone else has read the previous few pages about spinning down drives under ESXi, I have now realised why drives didn't seem to be spinning down - check out this post here re disabling ESXi's poll of SMART data from all drives every 30 mins!! ::)

 

http://www.avsforum.com/t/1442343/im-in-the-mood-for-some-renovation#post_22665051

Dude, you're killing me. No one spins down drives unders ESXi. Your using RDMs whether physical or virtual RDMs, the hypervisor owns the controller(s) and disk(s), thus the above is a hack, which needs to be thoroughly understood, that it will affect all drives including your datastore drives, so there goes your alerts that ESXi would have provided you with, had you not hacked it. If you want a guest OS to own disk(s) you have to use VMDirectPath and pass-through the entire controller to the guest OS. RDMs have very specific business case usage and it wasn't developed for unRAID nor spinning down drives  :P

Link to comment
madburg Collaborator

madburg

Posted
Posted

This is quite simple Tom, Everytime unRAID disk spindown setting is reached, in my case I have it set at 30 mins, any media that is playing has a hard freeze for a few seconds. If I disable the spindown setting the freeze does not occur. Afp, clients running plex.

 

As a secondary test I ran the script here and do not experience the freeze at all and the disks spin down as required.

Thank you for the summary, that's information I can use.  I will try and duplicate that.

Can a part of this thread please make itself back into the Issue List for tracking, as a clear long standing issue is mixed up with VMWare RDM stuff. Many thanks in advanced.

Link to comment
TheDragon Contributor

TheDragon

Posted
  • Author
Posted

Just in case anyone else has read the previous few pages about spinning down drives under ESXi, I have now realised why drives didn't seem to be spinning down - check out this post here re disabling ESXi's poll of SMART data from all drives every 30 mins!! ::)

 

http://www.avsforum.com/t/1442343/im-in-the-mood-for-some-renovation#post_22665051

Dude, you're killing me. No one spins down drives unders ESXi. Your using RDMs whether physical or virtual RDMs, the hypervisor owns the controller(s) and disk(s), thus the above is a hack, which needs to be thoroughly understood, that it will affect all drives including your datastore drives, so there goes your alerts that ESXi would have provided you with, had you not hacked it. If you want a guest OS to own disk(s) you have to use VMDirectPath and pass-through the entire controller to the guest OS. RDMs have very specific business case usage and it wasn't developed for unRAID nor spinning down drives  :P

I do, and the person in the thread I linked to does!!  :P  I'm aware its hardly the ideal situation, but aside from unRAID drives, I currently only have 1 datastore drive - which is the only drive that would be effected by doing this.  It's an old drive I had going spare so I'm not too worried about that, and I have backed up the contents anyway. My original plan was to pass-through my entire controller, but sadly despite both my motherboard and cpu supposedly 'supporting' VT-d, for some reason they don't play nice with ESXi.  I considered at the time, whether I could 'bodge' it for the time being, and RDM all the drives - but spindown and temperatures didn't work at all back then.  That's why when Weebo mentioned that things may have changed in the announcement thread, I decided I'd test it out and essentially have another crack at it - eg. figure out how to get the UPS shutting down the VMs and the host cleanly etc.  Really it's only a temporary measure until I upgrade my hardware, and letting me have a tinker with ESXi in the meantime ;)

Link to comment
madburg Collaborator

madburg

Posted
Posted

Just in case anyone else has read the previous few pages about spinning down drives under ESXi, I have now realised why drives didn't seem to be spinning down - check out this post here re disabling ESXi's poll of SMART data from all drives every 30 mins!! ::)

 

http://www.avsforum.com/t/1442343/im-in-the-mood-for-some-renovation#post_22665051

Dude, you're killing me. No one spins down drives unders ESXi. Your using RDMs whether physical or virtual RDMs, the hypervisor owns the controller(s) and disk(s), thus the above is a hack, which needs to be thoroughly understood, that it will affect all drives including your datastore drives, so there goes your alerts that ESXi would have provided you with, had you not hacked it. If you want a guest OS to own disk(s) you have to use VMDirectPath and pass-through the entire controller to the guest OS. RDMs have very specific business case usage and it wasn't developed for unRAID nor spinning down drives  :P

I do, and the person in the thread I linked to does!!  :P  I'm aware its hardly the ideal situation, but aside from unRAID drives, I currently only have 1 datastore drive - which is the only drive that would be effected by doing this.  It's an old drive I had going spare so I'm not too worried about that, and I have backed up the contents anyway. My original plan was to pass-through my entire controller, but sadly despite both my motherboard and cpu supposedly 'supporting' VT-d, for some reason they don't play nice with ESXi.  I considered at the time, whether I could 'bodge' it for the time being, and RDM all the drives - but spindown and temperatures didn't work at all back then.  That's why when Weebo mentioned that things may have changed in the announcement thread, I decided I'd test it out and essentially have another crack at it - eg. figure out how to get the UPS shutting down the VMs and the host cleanly etc.  Really it's only a temporary measure until I upgrade my hardware, and letting me have a tinker with ESXi in the meantime ;)

@jack0w, its just that you didn't mention these points and your posting in a forum where someone may end up trying and not know the repercussions.

Weebotech has at many times stepped in and verified others scripts and cleared up misunderstandings (as have others), people need to know the whole picture before making a decision, your post lacked that details, thus my post. If it doesn't bother you, more power to you, but all this is not a solution for a production system and others should be aware. Tinkering, test bed, etc. all fine and dandy.

 

P.S. no guarantee that the CIM components or other parts in esxi don't depend on getting back SMART data from drives don't cause other issues. This is very grey->black area doing such a hack.

Link to comment
TheDragon Contributor

TheDragon

Posted
  • Author
Posted

@jack0w, its just that you didn't mention these points and your posting in a forum where someone may end up trying and not know the repercussions.

Weebotech has at many times stepped in and verified others scripts and cleared up misunderstandings (as have others), people need to know the whole picture before making a decision, your post lacked that details, thus my post. If it doesn't bother you, more power to you, but all this is not a solution for a production system and others should be aware. Tinkering, test bed, etc. all fine and dandy.

 

P.S. no guarantee that the CIM components or other parts in esxi don't depend on getting back SMART data from drives don't cause other issues. This is very grey->black area doing such a hack.

 

I can only offer my apologies, as I certainly didn't mean to offend you or anyone else.  I was just trying to give a complete picture of the situation to anyone else happens to be reading this thread, who is intending to try what I did. I honestly hadn't considered that anyone on the forum would just run commands they see people mentioning without knowing what the effects would be, and preparing themselves accordingly.

 

I had thought that the information I provided may prove useful to many of the people who are running ESXi, as there are plenty of people on these forums using ESXi with RDM drives in unRAID.

 

In the spirit of giving further detail, here's some more info on SMART monitoring and the smartd daemon http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2040405

 

Once again, many apologies - I didn't mean to offend, hurt, or annoy anyone with my post. I was just trying to be helpful to others.  :-[

 

 

Link to comment
madburg Collaborator

madburg

Posted
Posted

@jack0w, its just that you didn't mention these points and your posting in a forum where someone may end up trying and not know the repercussions.

Weebotech has at many times stepped in and verified others scripts and cleared up misunderstandings (as have others), people need to know the whole picture before making a decision, your post lacked that details, thus my post. If it doesn't bother you, more power to you, but all this is not a solution for a production system and others should be aware. Tinkering, test bed, etc. all fine and dandy.

 

P.S. no guarantee that the CIM components or other parts in esxi don't depend on getting back SMART data from drives don't cause other issues. This is very grey->black area doing such a hack.

 

I can only offer my apologies, as I certainly didn't mean to offend you or anyone else.  I was just trying to give a complete picture of the situation to anyone else happens to be reading this thread, who is intending to try what I did. I honestly hadn't considered that anyone on the forum would just run commands they see people mentioning without knowing what the effects would be, and preparing themselves accordingly.

 

I had thought that the information I provided may prove useful to many of the people who are running ESXi, as there are plenty of people on these forums using ESXi with RDM drives in unRAID.

 

In the spirit of giving further detail, here's some more info on SMART monitoring and the smartd daemon http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2040405

 

Once again, many apologies - I didn't mean to offend, hurt, or annoy anyone with my post. I was just trying to be helpful to others.  :-[

No apologies necessary, no one is offended, we just want everyone to be aware of the details of such a change.

All good  8)

Link to comment
jaybee Contributor

jaybee

Posted
Posted

OK read the thread now, admitedly skim read. Would it be fair to say that there are two issues in here that for some reason wondered off from the RC16C thread.

 

1: Freezes during movie playback if a disk spins down

2: ESXi issues with spindown

 

With number 2... that's surely not something that should be delaying a 5 release? i.e. How XYZ works under ESXi?

With 1, sounds like some investigation is required.

Link to comment
TheDragon Contributor

TheDragon

Posted
  • Author
Posted

OK read the thread now, admitedly skim read. Would it be fair to say that there are two issues in here that for some reason wondered off from the RC16C thread.

 

1: Freezes during movie playback if a disk spins down

2: ESXi issues with spindown

 

With number 2... that's surely not something that should be delaying a 5 release? i.e. How XYZ works under ESXi?

With 1, sounds like some investigation is required.

 

#1 I believe Tom is investigating that.

#2 You're right, it isn't an unRAID problem, and it isn't delaying the release of version 5.

 

Hope that very brief summary helps!  ;)

Link to comment
madburg Collaborator

madburg

Posted
Posted

OK read the thread now, admitedly skim read. Would it be fair to say that there are two issues in here that for some reason wondered off from the RC16C thread.

 

1: Freezes during movie playback if a disk spins down

2: ESXi issues with spindown

 

With number 2... that's surely not something that should be delaying a 5 release? i.e. How XYZ works under ESXi?

With 1, sounds like some investigation is required.

correction #2 = ESXi utilizing RDM, issues with spindown

as VMDirectPath works just fine.

 

Nothing is delaying final, Tom knows just how much or little before he stamps final. We can all ask and point things out but he decides.

 

Link to comment

Archived

This topic is now archived and is closed to further replies.

Go to topic listing
×
×
  • Create New...