Jump to content
XPEnology Community

Enable port multiplier


janzy

Recommended Posts

  • 2 weeks later...

First I did the bios hack and then installed a Highpoint Rocket 622 eSata port pci express 2.0x1 controller card.

I have the eSata cable going from one of the card ports to a SansDigital TR4M+BNC Tower Raid 4 eSata tower.

Working great.

Link to comment
Share on other sites

I have two synology dx-510 enclosures left orphan after sudden death of 1510+ motherboard. So I tryied for two months to find any solution that supports esata's PM with those dx-510. I tried all internal controllers on 4 different motherboards, hp n54l and something up to 10 different types of external pci-e controllers. Those experiments included chipset-based e-sata (no luck at all) and chips form marvell, asmedia and silicon image.

 

So briefly let me not mention those controllers that do not support PM at all, those that do not support needed type of PM (sorry, I don't remember exact name - smth about command-based PM) and those that saw 1-2-4 disks from those 5 hdd-s situated in dx-510. Also I'll left apart the one which saw only 5 hdd's insteed of 10 from two dx-510. :smile:

 

The first lucky experience was with HPT RR 622. It worked with all my 10 hdd-s (after manual driver compilation from hpt sources). But... It suffered some sort of lockups under high disk transfer. For example while writing to array via nfs. Sometimes it happened after 10 minutes and on the other hand it may happen after 10 hours. It depended on cables (tried 3 different one), motherboards (4 attemps) and psu (3 test items). Also it depended on weather on Mars... Also it may happen on reading or array creation. But the result was always constant - the controller lockup happened sooner or later. At last I tried it in comparable situation with raid5 on suse 13.1. It worked for nearly a day... But at last failed. So I looked through google and found on ubuntu forum:

http://ubuntuforums.org/showthread.php?t=1592227

 

This thread is lasting for more than 3 years. I looked it through. There are no any reasonable resolve of the issue.

 

As for HPT R 622 (I mean Rocket not RocketRaid) - I do not know. I haven't tried this variant ('cause I do not know how to reflash its firmware from raid to non-raid).

 

Reading that thread I found a solution. Not with HPT RR 622, but for my problem. I purchased sil3132 based card. And it is really working without problems in linux. But there is a problem in xpenology. It does not work with both 4.3 and both 5.0 boot images (from Trantor and Gnoboot). Controller's driver crashes after turning dx-510 on while running xpenology, or lockups while booting if it was turned on before boot. I asked both authors on the situation - but nothing changed till now.

 

There are two solutions. First: I can use ESXi and install 4.3 or 5.0 as a VM. Second: I can use 4.2 (tried 3211-1.2.7 from Trantor). The second variant works fine except the well-known problem with recreating system partion's raid1 after reboot. Both variants were checked on hp n54l and ordinal asus 1155 mobo. Non-working with 4.3 and 5.0 was checked with latest firmwares for sil3132 (both raid and non-raid).

Link to comment
Share on other sites

First I did the bios hack and then installed a Highpoint Rocket 622 eSata port pci express 2.0x1 controller card.

I have the eSata cable going from one of the card ports to a SansDigital TR4M+BNC Tower Raid 4 eSata tower.

Working great.

 

The RocketRaid 622 card (comes with the 5-bay SansDigital TR5M+B esata tower) did not work in my N40L + Xpenology. From what I have read rocket 622 (not RR) does not even have port multipliers, at lease not without diver hacking.

 

Would be nice to have N40L's own esata PM work. I got some hints (described in this post: viewtopic.php?f=2&t=2272#p11931). but really need to read more to understand how it might work.

Link to comment
Share on other sites

First I did the bios hack and then installed a Highpoint Rocket 622 eSata port pci express 2.0x1 controller card.

I have the eSata cable going from one of the card ports to a SansDigital TR4M+BNC Tower Raid 4 eSata tower.

Working great.

 

The RocketRaid 622 card (comes with the 5-bay SansDigital TR5M+B esata tower) did not work in my N40L + Xpenology. From what I have read rocket 622 (not RR) does not even have port multipliers, at lease not without diver hacking.

 

Would be nice to have N40L's own esata PM work. I got some hints (described in this post: viewtopic.php?f=2&t=2272#p11931). but really need to read more to understand how it might work.

Link to comment
Share on other sites

The RocketRaid 622 card (comes with the 5-bay SansDigital TR5M+B esata tower) did not work in my N40L + Xpenology. From what I have read rocket 622 (not RR) does not even have port multipliers, at lease not without diver hacking.

 

Would be nice to have N40L's own esata PM work. I got some hints (described in this post: viewtopic.php?f=2&t=2272#p11931). but really need to read more to understand how it might work.

 

RR 622 needs drivers. You can see this page:

http://highpoint-tech.com/USA_new/serie ... wnload.htm

 

It should be compiled against running kernel sources. The same situation also in other linux distros. After building the module it should be insmoded and added somehow to initialization sequence of xpenology.

 

As for n40l/n56l internal e-sata port I am sure that it does not support PM FIS–based switching even with any hacked bios version. (I was mistaken in my previous message here - we need not command-based switching but FIS–based switching.)

 

Have a look at this table:

https://ata.wiki.kernel.org/index.php/S ... e_features

 

Here we should go to line with AMD SB820M as far as I see from HP's 13716_na.pdf. And at that raw we find out that PMP SM = CBS. So it will not work with FBS enclosures and as for CBS enclosures it will be very slow and useless for raid configuration.

 

CBS/FBS is not the only one limitation. There are quite a lot of reasons even for FBS controller card also not to work with the specific enclosure.

Link to comment
Share on other sites

The RocketRaid 622 card (comes with the 5-bay SansDigital TR5M+B esata tower) did not work in my N40L + Xpenology. From what I have read rocket 622 (not RR) does not even have port multipliers, at lease not without diver hacking.

 

Would be nice to have N40L's own esata PM work. I got some hints (described in this post: viewtopic.php?f=2&t=2272#p11931). but really need to read more to understand how it might work.

 

RR 622 needs drivers. You can see this page:

http://highpoint-tech.com/USA_new/serie ... wnload.htm

 

It should be compiled against running kernel sources. The same situation also in other linux distros. After building the module it should be insmoded and added somehow to initialization sequence of xpenology.

 

As for n40l/n56l internal e-sata port I am sure that it does not support PM FIS–based switching even with any hacked bios version. (I was mistaken in my previous message here - we need not command-based switching but FIS–based switching.)

 

Have a look at this table:

https://ata.wiki.kernel.org/index.php/S ... e_features

 

Here we should go to line with AMD SB820M as far as I see from HP's 13716_na.pdf. And at that raw we find out that PMP SM = CBS. So it will not work with FBS enclosures and as for CBS enclosures it will be very slow and useless for raid configuration.

 

CBS/FBS is not the only one limitation. There are quite a lot of reasons even for FBS controller card also not to work with the specific enclosure.

Link to comment
Share on other sites

Thanks for detailed post. Very good information

 

It should be compiled against running kernel sources. The same situation also in other linux distros. After building the module it should be insmoded and added somehow to initialization sequence of xpenology.

I need to learn how to compile the driver for RR622

 

As for n40l/n56l internal e-sata port I am sure that it does not support PM FIS–based switching even with any hacked bios version. (I was mistaken in my previous message here - we need not command-based switching but FIS–based switching.)

...

CBS/FBS is not the only one limitation. There are quite a lot of reasons even for FBS controller card also not to work with the specific enclosure.

 

Before using Xpenology, I was running win8 on BIOS hacked N40L. The tower worked with the onboard esata port. So in theory, it could work. Performance is a different story though.

Link to comment
Share on other sites

Thanks for detailed post. Very good information

 

It should be compiled against running kernel sources. The same situation also in other linux distros. After building the module it should be insmoded and added somehow to initialization sequence of xpenology.

I need to learn how to compile the driver for RR622

 

As for n40l/n56l internal e-sata port I am sure that it does not support PM FIS–based switching even with any hacked bios version. (I was mistaken in my previous message here - we need not command-based switching but FIS–based switching.)

...

CBS/FBS is not the only one limitation. There are quite a lot of reasons even for FBS controller card also not to work with the specific enclosure.

 

Before using Xpenology, I was running win8 on BIOS hacked N40L. The tower worked with the onboard esata port. So in theory, it could work. Performance is a different story though.

Link to comment
Share on other sites

Hi Deynet-

 

I think I have he same problem as you, I cannot get my sil3132 card to properly work with port multiplier, but it is detected (only 1 of 4 drives detected). Is this what you are seeing too?

 

Strangely, I can't get mine to do anything different on 4.2 or 4.3. Only shows 1 drive under either build. However, if I boot to Ubuntu, all the drives show up, so I know the hardware is good.

Link to comment
Share on other sites

Hi Deynet-

 

I think I have he same problem as you, I cannot get my sil3132 card to properly work with port multiplier, but it is detected (only 1 of 4 drives detected). Is this what you are seeing too?

 

Strangely, I can't get mine to do anything different on 4.2 or 4.3. Only shows 1 drive under either build. However, if I boot to Ubuntu, all the drives show up, so I know the hardware is good.

Link to comment
Share on other sites

  • 6 months later...
  • 2 months later...
  • 3 months later...

to update, I've got my tr4m+bnc now attached to my n54l through this usb3/esata adapter that supports port mulitplication.

 

http://www.amazon.com/gp/product/B005DC ... UTF8&psc=1

 

The drives show up individually as usb devices but they don't present themselves to the storage manager, perhaps because they are external? I'll have to play around with it and see if I can find a config that will work for backup purposes. If I can't get a way to span the drives into 1 volume, maybe I can configure the backups to simply go to the next one as one fills up - not sure if there is some sort of JBOD functionality to the backup system that supports mulitple disks. I'll chip away at it.

 

this link may come in handy if I want to try to force feed things:

http://forum.synology.com/wiki/index.ph ... AID1_array

Link to comment
Share on other sites

I am able to use the N54L onboard eSATA's port multiplier connected to my Cineraid 4 bay right out of the box with Xenserver used as a hypervisor. I do have the bios mod installed as well.

 

I'm using Xenserver 6.5, and have successfully added 4 additional local disks from my Cineraid box to my Xenserver base install. Xpenology 5.1 is loaded as a VM and can utilize up to 2 of these disks right now. The main problem I am having right now is since the N54L does not support IOMMU I can't pass-through the sata disks right to Xpenology, which I posted about in another thread:

 

viewtopic.php?f=2&t=5409

 

Getting around this current pass-through limitation I think would set everyone home free to use the onboard eSATA port.....

Link to comment
Share on other sites

×
×
  • Create New...