LSI SAS HBA card Info for RS18016xs+

[Synology_Service]

I truly would appreciate any help on this.

I seriously mean it too.

I have a dilemma.

And this is a off shoot question.

I have been battling this for 2 months now.

I can't find anything on the web for the RS181016xs SAS HBA card that is in it.

I did see someone mention it is a LSI card.

But that was it.

If anyone could help and post more details about that card.

 

What's also odd.

12 Disk Servers and NAS's are common all over the world.

But for some reason. You can't find a 12 port(disk) card.

And Synology has 12 port(disk) cards in its servers.

3 Connection points on all their cards. Each one supports 4 drives. ===12 Disks

 

As an example I have attached the SAS HBA card from a 18017xs+

Ad yes. You can see the 3 connecting ports on it.

But you won't find that style on the internet.

I have tried.

 

And I tested this card in the RS18016xs+. And it didn't work for some reason, as its from a RS18017xs+

And you would think they would have at least a drive for this card too on the system.

Synology has this card in stock for $800. LOL! That's insane!

 

 

[smileyworld]

I can't help you with this exact HBA, but what I do know is that my two LSI SAS 9211-8i (HBA in IT mode) work perfectly.

 

If you are interested I can give you more insight in a month with the Broadcom SAS 9300-16i.

 

Did you edit your user_config.json accordingly?

 

Tutorial how to do more than 12 drives:

 

 

[smileyworld]

Quick follow up on this.

I switched from 2x LSI SAS 9211-8i (HBA in IT mode) to 1x Broadcom SAS 9300-16i and it works without any issues.

[Synology_Service]

So these cards work in the 18106xs+ You had to make no changes, and the OS supported the drivers? Synology hinted it was a LSI card. Just the name wasn't on the card that way. Just a Synology part number. But when speaking to support there, they had mentioned this.

So these do work on that model for sure?

[IG-88]

i do remember this card

https://xpenology.com/forum/topic/56872-develop-and-refine-the-ds3622xs-loader/page/8/#findComment-266961

 

the card is most likely a broadcom chip based, custom synology card, never seen any 12i4e cards anywhere else, even 8i8e is exotic,, the chip will be a 16port broadcom you will find on 16i cards

not sure how much custom the firmware might be and there is a possibility that DSM has some special moves ready for specific cards with synologys own firmware

i do remember special code in syno's modded kernel source for activation of sata multiplexing for a special a chip with special firmware, so there might be also such things for syno's own sas cards

you could dig through the source and check the update files for these models to see if there is any special firmware files for these controllers (they might get updates from time to time and they would need to be in synologys dsm *.pat files) - that would be a lot of work and might lead to no solution of your problem, most likely it will take some time and you gain some knowledge

 

a simple solution would be to check the chip on the original card and the replacement card you tried by removing the cooler, if they are different then this could be the reason the card from a newer model did not work in the older hardware, if they are the same then there might be firmware differences (most simple case might be a version did not match)

best way would be to check if the spare parts synology offers on there homepage are the same or not

my guess would be that they are the same part and there are just problems with firmware versions, i dont think they will have the kind of sophisticated firmware detection and upgrade capabilities 3par has on its san systems so you might need to match these things manually and doing that for oem synology controllers you have no easy access to firmware files and tools

you could dig into dsm's *.pat files and the installed system to find these things and might be able to manually do things but even this is quiet some work and when you brick the controller with improper firmware or procedure ...

 

the comments above will most likely reference to "oem hardware running dsm" aka a xpenology system using a special modded loader that also supplies additional drivers

you can try to stick in some "older" random lsi/avago/broadcom card with it-mode firmware but the outcome is open and you not have any disks with data you need in these test system

its not likely anything problematic with the hardware will happen

when using a 8i controller mentioned above it might fill out 8 of the 12 possible ports and a 16i might use all 12 ports but it might also not work at all, so dont shell out money as it might not lead you to a solution - also check the cables these older 9211-8i have different connectors, might be best to use a newer 8i/16i that comes with the connectors uses in the system like a 9311-8i

 

 

 

Edited March 1 by IG-88

[Synology_Service]

Thanks for the reply. And I agree. My hopes were someone had this card in their system, and was able to see the chipset itself. I was only told many are LSI rebranded cards. And yes about the cables.

In one case of changing the card from a 181017xs+ chassis over to this 18016xs+ chassis, the SFF connectors are different. I had a set of adapters though for testing. And to no avail with the card I had tested with.

I know this is a non Xpenology related question. Was just hoping in the back ground that someone had this unit, and a Xepenology from the forum, and would relate. Appreciate the answers for sure. And Thanks!--Ray

[Synology_Service]

Something interesting. Happen to be digging in the 181016xs+ PAT file for any info on the card or cards it might use. Still have alot more digging to do. ANd trying to see if I spot a driver maybe in there for the SAS card type. But how strange to see the requests for a Dell Megaraid H700 PERC card as a SAS controller card. Those I know can run in HBA and SAS, like the OEM Synology ones do. Makes me wonder if the supplier for Dell cards might be what Synology is using for this model NAS and its SAS HBA Card.

Or also spotted as shown for 3WARE cards, 

See pic of a area I highlighted in the PAT file.

 

Edited March 1 by Synology_Service

[IG-88]

2 hours ago, Synology_Service said:

Something interesting. Happen to be digging in the 181016xs+ PAT file

 

so 7.x pat files need to be extracted, did you do that? my guess is no because you do not mention any file name

extracted it would look like this and hda1.tgz would be the content of the dms partition

 

in the pat there can be also firmware updates, in this case its sas expander firmware

firmware\syno_sas_exp_fw\RX2417sas\

things like that i had in mind, there are some things that would need to run early to do bios and firmware updates (maybe some hints to that in rd.gz?)

 

tat would be some snipet ffrom the mpt3sas.ko driver, ans there are references to all kinds of controllers with that broadcom chip

 

 

 

so the part you have there might just be some random linux kernel code

there is a bunch of different controllers, not just broadcom, cciss is hpe specific, hpt (aka high point), areca, aacraid, ... all completly different stuff that need extra kernel drivers and these drivers are not part of dsm, there are usually ony drivers included that are needed for all models of the cpu type/class

i would suggest staring with rd.gz (extract) and look if there are some brachings for firmware update and then in hda1.gz (files like /etc/rc, rc.sas)

things like this would be catchy

fwupgrade

fw_upgrade

 

understanding the firmware update process will also help understand about types of controllers, version checks, ...

rd.gz extracted

\usr\syno\sbin\

syno_oob_fw_upgrade

sas_fw_upgrade.sh
->

#!/bin/sh

source /usr/syno/share/rootdevice.sh
source /usr/syno/share/synorbd.sh

if [ $# -lt 1 ]; then
	echo "Mount point should be passed as argument"
	return 0
fi

#Return value
RetSuccess=0
RetFail=1
RetFileCorrupt=2

Mnt=$1
SasFwDirName="sas_fw_upgrade_tool"
SasFwToolPosition="/tmp/$SasFwDirName"
FwUpgradeRecordFile="${Mnt}/$SasFwDirName/.fw_upgraded_record"
FwUpgradeLogFile="${Mnt}/$SasFwDirName/.fw_upgraded_log"
FwUpgradeTmpLogFile="/tmp/.fw_upgraded_log"
FwCheckScript="$SasFwToolPosition/check_fw.sh"
FwUpgradeScript="$SasFwToolPosition/upgrade_fw.sh"

KeySN="sn"
KeyRetry="retry"

RetryCount=0
RetryMax=3
CurSerial=`cat /proc/sys/kernel/syno_serial`

clear_retry_record()
{
	if [ -f $FwUpgradeRecordFile ]; then
		rm $FwUpgradeRecordFile
	fi
}

fw_upd_log()
{
	echo $1
	if [ -d "`dirname $FwUpgradeLogFile`" ]; then
		echo "[`date`] "$1 >> $FwUpgradeLogFile
	fi
}

# Check if upgrade tool there
if [ ! -d ${Mnt}/$SasFwDirName ]; then
	fw_upd_log "No SAS firmware upgrade tools"
	clear_retry_record
	return $RetSuccess
fi

fw_upd_log "Found SAS Firmware upgrade tool."
cp -rf ${Mnt}/$SasFwDirName $SasFwToolPosition

if [ ! -f $FwCheckScript ]; then
	fw_upd_log "Missing $FwCheckScript. Cancel SAS firmware upgrade."
	return $RetFileCorrupt
fi
if [ ! -f $FwUpgradeScript ]; then
	fw_upd_log "Missing $FwUpgradeScript. Cancel SAS firmware upgrade."
	return $RetFileCorrupt
fi

/bin/sh $FwCheckScript
ret="$?"
if [ $ret == "0" ]; then
	fw_upd_log "Already the newest version."
	clear_retry_record
	rm -rf ${Mnt}/$SasFwDirName
	return $RetSuccess
elif [ $ret == "2" ];then
	fw_upd_log "Firmware checksum mismatch."
	return $RetFileCorrupt
fi

if [ -f $FwUpgradeRecordFile ]; then
	OriSerial=`get_key_value $FwUpgradeRecordFile $KeySN`
	if [ "$OriSerial" !=  "$CurSerial" ]; then
		fw_upd_log "Mismatch host SN, skiping sas firmware upgrade"
		return $RetSuccess
	fi

	RetryCount=`get_key_value $FwUpgradeRecordFile $KeyRetry`
	if [ $RetryCount -ge $RetryMax ]; then
		fw_upd_log "Retry count reach $RetryMax times. SAS firmware upgrade FAIL!"
		return $RetFail
	fi
	fw_upd_log "Has upgraded $RetryCount times"
fi

fw_upd_log "SAS firmware version is different. Prepare to upgrade."
# Record the retry count and SN
RetryCount=$(($RetryCount + 1))
echo "$KeyRetry=\"$RetryCount\"" > $FwUpgradeRecordFile
echo "$KeySN=\"$CurSerial\"" >> $FwUpgradeRecordFile

# Umount $RootDevice and md1 then stop them
fw_upd_log "Swap Stop md1"
FailtoStopMd=1
for i in `seq 1 3`
do
	/usr/syno/bin/syno_swap_ctl --stop > $FwUpgradeTmpLogFile 2>&1
	if [ $? -ne 0 ]; then
		fw_upd_log "Swap stop md1 failed"
		fw_upd_log "`cat $FwUpgradeTmpLogFile`"
		sleep 2
		continue
	fi
	FailtoStopMd=0
	break
done
if [ $FailtoStopMd -eq 1 ]; then
	sync
	/sbin/reboot -f
fi
fw_upd_log "Swap stop md1 finished"

FailtoStopMd=1
fw_upd_log "Unmount $RootDevice"
if [ ! -z "`mount | grep $RootDevice`" ];then
	for i in `seq 1 3`
	do
		umount -f $RootDevice > $FwUpgradeTmpLogFile 2>&1
		if [ $? -ne 0 ]; then
			fw_upd_log "Unmount $RootDevice failed"
			fw_upd_log "`cat $FwUpgradeTmpLogFile`"
			sleep 2
			continue
		fi
		FailtoStopMd=0
		break
	done
else
	FailtoStopMd=0
fi
if [ $FailtoStopMd -eq 1 ]; then
	sync
	/sbin/reboot -f
fi
# Since here, $RootDevice is no longer umounted
fw_upd_log "Unmount $RootDevice finished"

FailtoStopMd=1
fw_upd_log "Disassembling $RootDevice"
if [ -b $RootDevice ];then
	UnloadSynoRbd
	for i in `seq 1 3`
	do
		mdadm -S $RootDevice > $FwUpgradeTmpLogFile 2>&1
		if [ $? -ne 0 ]; then
			fw_upd_log "Disassembling $RootDevice failed"
			sleep 2
			continue
		fi
		FailtoStopMd=0
		break
	done
else
	FailtoStopMd=0
fi
if [ $FailtoStopMd -eq 1 ]; then
	#mount $RootDevice again to record the error log
	LoadSynoRbd
	/bin/mount $RootDevice ${Mnt}
	fw_upd_log "Disassembling $RootDevice failed"
	fw_upd_log "`cat $FwUpgradeTmpLogFile`"
	sync
	/sbin/reboot -f
fi
fw_upd_log "Disassembling $RootDevice finished"

# Run the upgrade script
/bin/sh $FwUpgradeScript
/sbin/reboot -f

 

files with fw in hda1

 

same for sas

 

things like this, but you should do that with the pat files of the two RS you are interested in

 

Edited March 2 by IG-88

[Synology_Service]

Decided to take apart the heat sink on the 181017xs+ card just to see what it has as a chip. Uses Avago Raid Card. Chip is S101--0a!0-00

[Synology_Service]

This is the DSM6.2.4 PAT file for the 181016xs+.

You can also open the pat in hex editor. And edit directly into the PAT before distro. We do it alot in the Syno Service side.

Also below is that 18016xs+ PAT opened.

 

[IG-88]

19 minutes ago, Synology_Service said:

Decided to take apart the heat sink on the 181017xs+ card just to see what it has as a chip. Uses Avago Raid Card. Chip is S101--0a!0-00

as expected a sas3216, the 32 is the gen of controller, the 16 is the possible amount of ports

mpt3sas.ko would be the driver

as a broadcom controller it would be SAS 9300-12i4e, but the does not exist and i've never seen any other controller with that split of ports or that card layout (so its most likely a synology specific model)

 

16 minutes ago, Synology_Service said:

This is the DSM6.2.4 PAT file for the 181016xs+.

maybe there are higher chances of seeing firmware update files in 7.1.and 7.2 as there is more time between hardware release and that dsm version

Edited March 2 by IG-88

[Synology_Service]

Yea. Avago/LSI card. So they were right. It is a LSI card. Now this is for the 181017xs+. Its the 181016xs+ card I m trying to find.

As I tried this card already in the 181016xs+ and it didn't work. So was trying to hunt the driver down. That pat file I posted weas the 181016xs+ PAT for DSM 6.2.4

[Synology_Service]

You were right it was a 16 port. My guess was sticking to 12port. But it really is a 16 port. Just no through hole connector for the remaining 4 drives.

[Synology_Service]

Pat from here:

https://www.synology.com/en-us/support/download/RS18016xs+?version=6.2#system

 

[IG-88]

at least 08-08sas1212 shows a search result, but does not help much

https://eshop.evernex.com/product/sas-card-12g/

[IG-88]

when looking at the back of a 18016xs+ it looks very different, more like some conversion cable is used that might come from a internal riser card wir a sas controller ?

 

 

18017xs+

 

[IG-88]

looks like we need to slowly puzzle things together

only picture of the inside of the RS18016xs+ was in the qick start guide and there was no rise und in the place of the sas card was some kind of metal plate/blending, no 2nd card so it must be at least a 16 port card, but if there are 2 x external sas and 12 internal sas disks it might also be a 24 port controller

 

 

 

the spec of the RS18016xs+ only has 6G nothing about 12G so it must be a older sas controller then in the 17+ model

maybe a lsi sas2xxx series chip

https://www.thomas-krenn.com/en/wiki/LSI_6Gb/s_SATA%2BSAS_PCIe_3.0_HBAs

maybe a LSI SAS 92xx card (8 port or 16 port with sas expander to get to 24 ports) like LSI SAS 9201-16i

or something like this

https://docs.broadcom.com/doc/12352181

 

synology might check for the right controller or at least pci device id and if thats the case a newer controller/chip might prevent the RS18016xs+ from starting as its would have the wrong pci id, so some older cheaper 6G controller might work

 

so no need to check for any firmware differences, the 2016 RS came with SAS2 aka 6G and the 2017 RS with SAS3 aka 12G, definitely different controllers

 

Edited March 2 by IG-88

[Synology_Service]

 

Youtube video of the inside of the 181016xs+

 

 

 

 

 

tube.

Edited March 2 by Synology_Service

[Synology_Service]

Weird. The SAS ports are opposite the heat sink side.

[IG-88]

totally custom sas board and as there are 3 x 4 ports on one side and 2 x 4 ports to external a 16port is not enough, not seen any 24 port chips for that old 6G gen from lsi

so it either it uses some pcie bridge and 2 x 8port chip or one 8/16 port chip plus sas expander chip

but in both cases a normal sas pci id would be visible to the pcie bus, the only way to get this replaced would be with two 8port 6G sas cards and even if it boots it will for sure screw up the preset layout of the disk slots

depending on the case it might be easier to replace the boot media with a usb and use xpenology with 2 x 8port sas controller or maybe just use any other nas system

 

that also explains the pricing of the replacement board, and that might be a good price

Edited March 2 by IG-88

[IG-88]

if the newer controller will work or not might also depend on the dsm version, could be that when its running on 7.2 that it might work

the drivers did have some alterations over time and both controllers might be in the old line apart seen from driver (mpt2sas vs mpt3sas) and later the code of mpt2sas was integrated into mpt3sas and that driver "emulates" the missing mpt2sas - in old 6.x dsm the older 15 model might not have the driver needed

this could be check with the drivers present in the system, if there is only a mpt2sas the the newer controller can not work because of the missing driver (mpt3sas)

 

edit: i checked dsm 6.2 and there is only mpt2sas, so a RS18016xs+ system on that level will not be able to use the newer  controller because of the missing driver for a sas3 capable controller (mpt3sas.ko)

 

edit2: nope, even 7.2 only has mpt2sas.ko so the newer controller will not work simply because of the missing driver

in theory it might be possible to fix that by taking a mpt3sas.ko from a 7.2 xpenology loader and load that on the RS18016xs+, might not work at when a older  mpt2sas is already loaded but if you have a running RS18016xs+, add the controller and insmod the mpt3sas.ko you will see if it loads

might not work on boot because the mpt2sas is in rd.gz and the mpt3sas is not so you would not be able to access the boot volume (aka dsm system partition after initial load of rd.gz) and afair the rd.gz checksum protected an dsm checks it for integrity ... dead end

so the best course of action might be to test a one or two sas2 controller like 9211-8i in it-mode to see if the RS18016xs+ comes up with that instead of the original controller, 2 x 8ports  would be enough to have all 12 ports working and with a adapter cable even one external sas2 would be possible - depends on how the dsm system reacts to the differences in sas ports, there might be predefined mappings for ports to arrange them in a way it matches the physical disk slot layout

 

 

Edited March 2 by IG-88

[Synology_Service]

In the driver "mpt2sas.ko".

If you open it in hex. Does it mention the PID/VID values? Or a controller name or model?

 

Now with Synology new or old DSM.

Must support the card installed for the model that comes from factory as this would have come.

So DSM 6.2-7.2 would have to support the card no matter what.

As the nas will never change cards from manufacture.

If it did. Its rare. But I have seen it from Synology. The ko driver would have to be present in the DSM. Which makes me wonder why drivers for Dell, LSI, RocketRaid, etc are present in the pat file. To support those chipsets.

I believe the card is custom. Only by inputs and outputs. Hardware of the chipset, and design Synology borrows from other makers. Or tells the makers to remove what's not needed on the card for cost savings. So even with one SFF connector removed can save 1000's when so many are made. A penny saved. More profit.

The card is a 16 port card though one connector for the SFF is removed as this is a 12 bay instead of a 16. And the 2 external ports are for the expansion rack.

The ports I don't worry about as its rare for anyone to even buy an expansion bay.

So i'm kinda back to trying and hunt this card down, and who makes it. Or at least find out what what the chipset it uses to get a compatible driver if you see what I mean. Really appreciate you helping on this. Hope one day to make it up for you. Like if you ever need NAS service. It would be on me for you. And Thanks!

 

[IG-88]

52 minutes ago, Synology_Service said:

In the driver "mpt2sas.ko".

If you open it in hex. Does it mention the PID/VID values? Or a controller name or model?

you could use modinfo for that but dsm does not come ootb with that, you eiter can use some editor or feed the ko file to some place where you can use modininfo

with a editor you look for "vermagic", the area with the device id's will be just before that

but in this case its clear because the namening of the driver makes clear what it is for, mpt = SAS

1, mpt2 =SAS2 (6G), mpt3 = SAS3 (12G) and then newer controller is 12G so its SAS3 and will be in mpt3sas.ko (i've done a lot driver compiling for dsm 6.x and  there was a lot going on with these drivers)

 

1 hour ago, Synology_Service said:

The ko driver would have to be present in the DSM. Which makes me wonder why drivers for Dell, LSI, RocketRaid, etc are present in the pat file. To support those chipsets.

if there is something in the stock kernel drivers they need then it will be present in dsm, but you will often see that drivers are not present if they are not needed

in this case the type or class (cpu wise) does not come with sas3 hardware so there is no driver present even when the kernel would have one, the kernel option was not selected

here you will see cpu and the "package arch" will be what i called type or class of device, its about the deault kernel configuration the devices will have and that defines the driver it will have by default

https://kb.synology.com/en-global/DSM/tutorial/What_kind_of_CPU_does_my_NAS_have

if you look here it is "platform"

https://archive.synology.com/download/ToolChain/toolkit/7.2

here too but you can also see the kernel version its based on

(in some cases synology adds newer "external" drivers if needed like mellanox or the sas mpt3 drivers because the outdated kernel they use does not make up with the hardware they sell or sometimes newer optional parts are added and need additional drivers)

 

1 hour ago, Synology_Service said:

The card is a 16 port card though one connector for the SFF is removed as this is a 12 bay instead of a 16. And the 2 external ports are for the expansion rack.

 

the newer card has 16 sas port, the older sas2 card has 20 sas ports, for cost savings with the newer sas3 system (2017) they used a available 16 port chip and as they needed 12 ports internal and only had 16 in that cost efficient design they scraped the rarely used 2nd external sas port and as the reference designs for das3 cards only had 16i and 8i8e but they needed 12i4i they made a still more cost efficient new custom card with just one chip instead of the (presumably) way more complex 20 port sas2 card they had before in the 2016 system

 

in theory one LSI SAS 9201-16i card  (https://www.ebay.de/itm/132013639256) 

might be able to replace the custom card (the driver should work with this one) but you would not be having ootb external ports and there also can arise cooling problems when the heat sink is facing the housing and the airflow is not good enough, may work when the 16 replacement card has a lower power consumption but my guess is that this kind of mounting is not officially supported (the holes in the bracket imply that is mandatory that air goes over the heat sink ans leaves through these holes)

it also can have side effects on that slot numbering, but it might work that would be up to testing

in this case it would also need testing if a disk failure (disk missing) and then a reboot and replacing the disks work with the slot assignment of the disks

we have had recovery cases here where the lsi/broadcom stock driver "closed" gaps in the numbering on reboot and a replacement disk ended up with a higher number like /dev/sda is missing and then the former sdb got that place as sda and so on , bat that can also be a problem of the xpenology loader (shimming disks?) or synology modded the added mpt sas drivers (they did not provide source code for the external drivers they used)

 

[Synology_Service]

Thanks for the responses here. Appreciate this. Also sorry for the delay. Had some things come up. I might just try that.  Buying the LSI 9201 card and see how that plays out.

Makes me wonder if using the 9300 series will work as well?

I also happen to search that card with Synology in the search and spotted this below: 

https://www.reddit.com/r/DataHoarder/comments/ckojga/can_the_synology_diskstation_expansion_unit_be/

 

Also digging more into this 18106xs+ PAT file I see the mention of 3WARE Escalade. Areca, Rocket Raid, and Highpoint SAS cards. Areca is mentioned the most. There is a mention of Marvell SAS. But I think that is the chipset for a card. As Marvell makes chips. Not cards.

All in all really not sure why those are in the DSM for the Kernal to load up?

 

Edited March 3 by Synology_Service

[Synology_Service]

Or maybe I can ask you. Is there a way to search the PAT file for the drives it has that will be installed with the DSM?

This is a copy  of the opened PAT file from this model.