RC15 - Waking from standby works, but Tower is unreachable.

I can get the Tower to power on from standby, but nothing is usable. I can not access shares and the webgui and plugins are not responding. I will disable my plugins and see how it goes.

It happened again, this time only with mySQL plugin and Simple features loaded. I need mySQL for XBMC. I will try without Simplefeatures installed now.

Do you have access to the console so that you can check if the network has been restarted correctly?

Has this system worked with S3 before?    There are a LOT of systems that simply don't work well with S3 mode.  Also, do you have any add-in cards (SATA controllers, etc.) on the system?

 

Do you have access to the console so that you can check if the network has been restarted correctly?

Many systems need to have the network interface reset upon waking, and others need the same for their video.  There is a helper program s2ram (or something like that) often used.

 

Some systems never wake up properly.  (I've never gotten myC2SEE MB to wake up once put to sleep at all)

 

Joe L.

Some systems never wake up properly.  (I've never gotten myC2SEE MB to wake up once put to sleep at all)

 

Not surprised.  As I noted above, MANY systems do not respond well to S3.  It seems that desktop chipsets just don't handle it as well as the mobile chipsets;  and even if you have a system where it works, add-on cards can put it right back in the "doesn't work" category.

 

For maximum reliability, I always suggest you simply don't use S3 .. with no video and spundown drives you can get modern systems down to reasonably low power draws [My Atom system idles at ~ 20w;  my older system at ~ 80w].    If you want less, just shut the system down and use a magic packet to turn it on via WOL.

 

This system has worked reliably with S3 for years. This is the first trouble I have had with it. I tried connecting to the console remotely with IPMI and that fails as well. I have to use the Supermicro IPMI webpage to initiate a system reset. The only thing that changed was the upgrade to RC15. I may have to roll back to RC12, which stupidly, I did not backup so I need to find it elsewhere (found it). WOL from off just won't be tolerable for the rest of the family after they have gotten used to the speed of waking from S3 to access their media.

WOL from off just won't be tolerable for the rest of the family after they have gotten used to the speed of waking from S3 to access their media.

 

Understand ... of course there is something even faster than waking from S3 

 

WOL from off just won't be tolerable for the rest of the family after they have gotten used to the speed of waking from S3 to access their media.

 

Understand ... of course there is something even faster than waking from S3 

I would use "S0" with the drives spun down.    Likely to be much faster. 

Drives do spin down now, but I would still prefer it to go to S3 like it used to. I know this is not officially supported. Still figured I should report it and at least log my tests to see what may be causing it.

Some systems never wake up properly.  (I've never gotten myC2SEE MB to wake up once put to sleep at all)

 

Not surprised.  As I noted above, MANY systems do not respond well to S3.  It seems that desktop chipsets just don't handle it as well as the mobile chipsets;  and even if you have a system where it works, add-on cards can put it right back in the "doesn't work" category.

 

For maximum reliability, I always suggest you simply don't use S3 .. with no video and spundown drives you can get modern systems down to reasonably low power draws [My Atom system idles at ~ 20w;  my older system at ~ 80w].    If you want less, just shut the system down and use a magic packet to turn it on via WOL.

I spent a lot of time and effort in the beginning to make the systems energy efficient when running - but as you know yourself, 20w is somehow close to the minimum you can achieve for small servers - and with a 24 drive system you'll be a lot above that even when disks are spun down.

Thus I also changed to use S3 - which imho is the perfect solution for home use - "idle" power consumption is down to 1.6w - and the system is available within a few seconds, e.g. when starting media to play - it is almost "transparent".

As JoeL and others said - it is sometimes tricky and does not work with all systems - I also have to use s2ram to reinitialize some components and I pulled most of my hair when I tried to figure how it works some years ago ... that's also the reason I reused my old setups when upgrading recently, as this is for me personally a very important feature - and unfortunateoly there is only little information available what works and what doesn't ...

With modern desktop hardware, S3 suspend really should not be that hard to pull off. The difference between mobile and desktop hardware is decreasing with every generation. All of my desktops, be it Windows, OS X, or Linux have no problems going into and out of S3.

With modern desktop hardware, S3 suspend really should not be that hard to pull off. The difference between mobile and desktop hardware is decreasing with every generation. All of my desktops, be it Windows, OS X, or Linux have no problems going into and out of S3.

 

In many cases the issue isn't the basic system; but add-on hardware ... e.g. in the UnRAID case that would be additional SATA/RAID cards, perhaps an add-in NIC card, etc.

 

In my case to make it simple, I have no add-in cards. Everything is built onto the Supermicro mobo.  Eitherway it worked before upgrading to RC15.

I had this problem when I tried to get sleep to work with my system too.

 

One question I have for the people suggesting its hardware related... why does it work flawlessly with windows on the same hardware?

I had this problem when I tried to get sleep to work with my system too.

 

One question I have for the people suggesting its hardware related... why does it work flawlessly with windows on the same hardware?

... because it's different OS, drivers, etc.

I have had the same - tested with windows, all fine, sleep, WOL, etc. - then with unraid/slackw - network adapter not properly reset after wakeup, video net properly reinitialized after wakeup, ....

Actually, many of the issues are still caused by combination of bios/drivers (not hw, if it does work in other environments) and there is often e.g. non properly implemented acpi tables causing issues - and then it depends on the OS and drivers to handle and/or workaround that.

I am sure, ppl with deeper knowledge could explain better ... but that's what I was told when I asked your question ;-)

I had this problem when I tried to get sleep to work with my system too.

 

One question I have for the people suggesting its hardware related... why does it work flawlessly with windows on the same hardware?

... because it's different OS, drivers, etc.

I have had the same - tested with windows, all fine, sleep, WOL, etc. - then with unraid/slackw - network adapter not properly reset after wakeup, video net properly reinitialized after wakeup, ....

Actually, many of the issues are still caused by combination of bios/drivers (not hw, if it does work in other environments) and there is often e.g. non properly implemented acpi tables causing issues - and then it depends on the OS and drivers to handle and/or workaround that.

I am sure, ppl with deeper knowledge could explain better ... but that's what I was told when I asked your question ;-)

 

The above covers most of the reasons.  The simple fact is that ACPI can be very "tricky" to get everything right.  The #1 requirement is BIOS support.  If your system works perfectly in Windows, then clearly it's got that.  The next requirement is that the OS interfaces correctly with the BIOS ACPI drivers, and that the appropriate hardware needed for "wakeup" is left in the right state (e.g. NIC, video card, etc.), and that those things that need re-initialized (typically the video card) are indeed properly re-initialized on wakeup.

 

Since your hardware works perfectly in S3 with Windows, then it CAN work that way with Linux.    There is apparently a tool you can add to UnRAID that will do the appropriate initialization on wakeup, but I don't recall the specifics ... search these forums and I'm sure you can find it.  [Joe L mentioned it above -- "s2ram" is what you're looking for]

Lat time I tired playing with s2ram was painful. I sure hope I don't have to do that again. For now I may just roll back to RC12, which worked beautifully. My guess is it has something to do with the new Linux kernel in RC15.

Lat time I tired playing with s2ram was painful. I sure hope I don't have to do that again. For now I may just roll back to RC12, which worked beautifully. My guess is it has something to do with the new Linux kernel in RC15.

 

While that very well may work, the "problem" is that you'll be stuck with a Release Candidate rather than running v5.0 ("final").    While that may not be an issue, there ARE things that have been fixed along the way ... and it's clearly easier to troubleshoot any future problems if you're using the actual released version of the software.

 

It IS possible that something that changes between RC15 and v5.0 could resolve this, but if it's kernel-related, I doubt it.    I'd at least spend a bit of time with s2ram and see if you can get it to work with RC15.

 

Yeah, I will. I just don't have that "spare" time right now. As I play with it, I'll log my success/failures here. It was nice when it worked and worked with SimpleFeature S3 plugin. Super simple.

I plan to follow this as well ... if you get it working nicely I may even play around with it as well.  I tend to avoid S3 due to its many issues;  but it WOULD be nice if I could drop my systems' standby power to a couple of watts !!

 

I wonder if the sleep method has just changed for RC15 as per this thread:

http://lime-technology.com/forum/index.php?topic=27843.0

 

Sent from my SPH-D710 using Tapatalk 2

 

Accessing /sys/power and then querying

cat state

:

I got the result of:

freeze and mem.

running:

echo -n "mem" > state

Allowed the system to enter standby. I allowed it to sleep for a couple moments and then sent a WOL request. The tower woke but was unreachable. The front power light just blinks as if it is still sleeping, but the fans are all running. My WOL app still sees the tower as not on.

 

Because I was curious, after I rebooted I also tried the "freeze" state.

echo -n "freeze" > state

When I entered that, the system entered the state it is in after I try to wake from S3 currently. The front light blinks and the fans are running. My WOL app sees the tower as off. So it kinda seems when my system is waking from the "mem" state is is going into the "freeze" state. At this point sending it multiple WOL packets results in nothing. Once again I had to initiate a reboot with IPMI.

I rolled back to RC12a and checked the same things under it. Querying cat state this time I only got:

mem

I did not get "freeze" listed as I did with RC15. So I sent the command to put it to sleep as I did earlier. When I sent the WOL packet, the tower woke perfectly fine. I think this is a kernel or driver issue.

VERY interesting in deed! Thanks for your reply in my other thread, it seems we're on the same side 

I thought I would post an update on my quest for sleep under vs 5

 

I loaded up the new 5.0.1 and this time I got another option when querying the power states.

cat /sys/power/state

Now replies with:

freeze mem standby

I just tested using standby and the system went into standby and woke with a magic packet and everything works as expected.

echo -n standby > /sys/power/state

I just did another testing using "mem" this time and it alos worked! Wooohooo!

echo -n mem > /sys/power/state