Jump to content
XPEnology Community

RedPill TinyCore Loader Installation Guide for DSM 7.1 BareMetal


Peter Suh

Recommended Posts

8 hours ago, daryl said:

Hi Peter,

 

Hope you are well. Question: is there a script in TCRP that is supposed to expand the storage to the limit of the USB? I use MacOS and the DD method. Now on an Asus Tuff B460M M-ATX board LGA1200 and when I try to build for a DS3622xs+ I see a sed message near the end of the menu.sh script indicating that 409 items failed to write due to no space left on device. Using an 8gb usb stick.  Guessing this may be the reason I am not able to start the system. Looks like kernel panic. 

 

Thank you

Daryl

 

I hope that the difficult situation in the company is well resolved.


I'm not even sure if you're currently using this image I shared last time.


https://github.com/PeterSuh-Q3/tinycore-redpill/releases/tag/v0.9.3.0-1


This is an image that has been deliberately resized and increased using the gparted program for the shortage of partitions.


I like to use the DD command in MacOS, but I've never seen partitions easily adjusted with just one script.


You have to use Gparted if necessary, which is not easy.

Link to comment
Share on other sites

Thank you for the information. I ended up having to reinstall several times last night. My array is in SHR2 Raid format. The 3622 and every other desirable intel based model does not support SHR. In the menu script late last night I saw DS1019+ and this worked better than the DS920+ as I had no drives showing. 

 

In the DS1019+ I have two NVME's installed and the BIOS sees both. In /dev I have both nvme0 and nvme1 but DSM only sees one of them. I still need to RTFM on the motherboard and see if one of the NVME's deactivates if a given sata port is in use. Would think it would populate dev tree if it was deactivated. Not sure yet. One thing that does look on the horizon is the marriage of synology and AMD based on model offerings. That and I am guessing device tree is on x86 too? 

 

I'll use your image going forward. This was hit and miss and with no room left on device from SED. 

Link to comment
Share on other sites

5 hours ago, daryl said:

Thank you for the information. I ended up having to reinstall several times last night. My array is in SHR2 Raid format. The 3622 and every other desirable intel based model does not support SHR. In the menu script late last night I saw DS1019+ and this worked better than the DS920+ as I had no drives showing. 

 

In the DS1019+ I have two NVME's installed and the BIOS sees both. In /dev I have both nvme0 and nvme1 but DSM only sees one of them. I still need to RTFM on the motherboard and see if one of the NVME's deactivates if a given sata port is in use. Would think it would populate dev tree if it was deactivated. Not sure yet. One thing that does look on the horizon is the marriage of synology and AMD based on model offerings. That and I am guessing device tree is on x86 too? 

 

I'll use your image going forward. This was hit and miss and with no room left on device from SED. 

 

DS1019+ is a None Device-Tree model.

So far, there are two platforms related to AMD in XPENOLOGY, v1000 and r1000.
Both of these platforms are Device-Tree based.

And, there is another INTEL GeminiLake based on Device-Tree.


If you are a developer, you can take a look at my nvme-cache driver processing logic.

 

https://github.com/PeterSuh-Q3/rp-ext/blob/main/nvme-cache/src/nvme-cache.sh#L54


It contains logic that automatically handles Hexa edit of libsynonvme.so.1 as a script.
Unfortunately, DS1019+ supports only one nvme in the libsynonvme.so.1 file.


Based on 2 NVMe and Device-Tree, the most suitable model with AMD would be DS923+.
It works perfectly in M SHELL for TCRP FRIEND.
Recently, one issue with this model was the inability to start power reserve.
This problem is also solved in my M SHELL.

 

I don't know if you ever mentioned the issue of HBA,
It does not work on Device-Tree base.
And it works on ApolloLake platforms such as DS918+ and DS1019+,

but there is a problem that the disk manager does not properly recognize the S.M.A.R.T disk serial number.

Edited by Peter Suh
Link to comment
Share on other sites

Hi Peter,

 

I bought Intel hardware after running into issues with the HBA's I have on v1000 platform. Think it was the DS1621. Ryzen5600G was what I replaced. My comments came from going through all the available models and seeing what looked to align. Seeing more AMD. Concern being that the best intel models synology offer either don't offer nvme or they don't support SHR. Do you know why this limitation on SHR exists on intel? 

 

My bad on device tree. I thought I saw it on the screen as build was taking place on a gemini lake I selected. Can't remember the model as I think I tried them all. All the intel at least. 

 

On my current hardware the smaller of my HBA's recognized but I noticed the drives running super hot so I have a new case coming. Rosewill hot swap cages not providing enough cooling. Will be transplanting the hardware to a new case next week and would like to select an intel model that can support SHR and dual NVME. A developer I am not. Thank you for the information. That would explain why I see two NVME listings but only one getting allocated.

 

Aside from DS1019+ what other models would be recommendable? I have Asus TUF B460M, I-7 - 10700, 32gb ram and using a dual nic intel card e1000. The HBA I willl use is a LSI00301 SAS 9207-8i. It gets picked up by bios and added to bottom in that I can use bios to change the HBA settings. Currently the LSI9212 is getting recognized with both being plugged in. I don't think the 8 port card is getting recognized however my disk numbering starts at 9 so I am not confident on that. 

 

DS920+ also looked close but I wasn't able to get any HDD's to recognize. Saw it mentioned in this thread but found it faster to try a different model than to try to follow along to see if work around came along. 

Link to comment
Share on other sites

3 hours ago, daryl said:

Hi Peter,

 

I bought Intel hardware after running into issues with the HBA's I have on v1000 platform. Think it was the DS1621. Ryzen5600G was what I replaced. My comments came from going through all the available models and seeing what looked to align. Seeing more AMD. Concern being that the best intel models synology offer either don't offer nvme or they don't support SHR. Do you know why this limitation on SHR exists on intel? 

 

My bad on device tree. I thought I saw it on the screen as build was taking place on a gemini lake I selected. Can't remember the model as I think I tried them all. All the intel at least. 

 

On my current hardware the smaller of my HBA's recognized but I noticed the drives running super hot so I have a new case coming. Rosewill hot swap cages not providing enough cooling. Will be transplanting the hardware to a new case next week and would like to select an intel model that can support SHR and dual NVME. A developer I am not. Thank you for the information. That would explain why I see two NVME listings but only one getting allocated.

 

Aside from DS1019+ what other models would be recommendable? I have Asus TUF B460M, I-7 - 10700, 32gb ram and using a dual nic intel card e1000. The HBA I willl use is a LSI00301 SAS 9207-8i. It gets picked up by bios and added to bottom in that I can use bios to change the HBA settings. Currently the LSI9212 is getting recognized with both being plugged in. I don't think the 8 port card is getting recognized however my disk numbering starts at 9 so I am not confident on that. 

 

DS920+ also looked close but I wasn't able to get any HDD's to recognize. Saw it mentioned in this thread but found it faster to try a different model than to try to follow along to see if work around came along. 

 

As I already mentioned, to use the HBA you must select the None Device-Tree platform.
The most stable platform with the HBA is the Broadwellnk DS3622xs+.
Two NVMe caches are also supported reliably.

 

Don't worry about the lack of official support for SHR on DS3622xs+.
Models that support SHR are mainly the home target DS918+.
On the DS3622xs+, which is an enterprise target, support for SHR was dropped midway through revisions.
What was possible in DSM 7.0.1 probably stopped in DSM 7.1.

 

I am currently using DS3622xs+ and SHR.
However, it was possible because I migrated from DSM6 DS918+ where I was already using SHR.


Creating SHR volumes from scratch on DS3622xs+ is not possible by default.
There is a way to re-enable SHR by adjusting the Synology environment file.
This method is easy to find with a little Googling.


The above method is recommended because there is no model where SHR and HBA are stable at the same time.

Edited by Peter Suh
Link to comment
Share on other sites

16 hours ago, daryl said:

Hi Peter,

 

I bought Intel hardware after running into issues with the HBA's I have on v1000 platform. Think it was the DS1621. Ryzen5600G was what I replaced. My comments came from going through all the available models and seeing what looked to align. Seeing more AMD. Concern being that the best intel models synology offer either don't offer nvme or they don't support SHR. Do you know why this limitation on SHR exists on intel? 

 

My bad on device tree. I thought I saw it on the screen as build was taking place on a gemini lake I selected. Can't remember the model as I think I tried them all. All the intel at least. 

 

On my current hardware the smaller of my HBA's recognized but I noticed the drives running super hot so I have a new case coming. Rosewill hot swap cages not providing enough cooling. Will be transplanting the hardware to a new case next week and would like to select an intel model that can support SHR and dual NVME. A developer I am not. Thank you for the information. That would explain why I see two NVME listings but only one getting allocated.

 

Aside from DS1019+ what other models would be recommendable? I have Asus TUF B460M, I-7 - 10700, 32gb ram and using a dual nic intel card e1000. The HBA I willl use is a LSI00301 SAS 9207-8i. It gets picked up by bios and added to bottom in that I can use bios to change the HBA settings. Currently the LSI9212 is getting recognized with both being plugged in. I don't think the 8 port card is getting recognized however my disk numbering starts at 9 so I am not confident on that. 

 

DS920+ also looked close but I wasn't able to get any HDD's to recognize. Saw it mentioned in this thread but found it faster to try a different model than to try to follow along to see if work around came along. 

 

This is a method of forcibly enabling SHR on models that do not officially support it.

 

https://www.doyler.net/security-not-included/enable-shr-on-synology-ds3615xs

Link to comment
Share on other sites

Hey Peter,

 

Amazing. Very helpful. Thank you! Never thought to look at adding SHR support. You have given me a weekend project. At a point of low confidence I was researching raid options I saw SnapRaid and wondered your opinion on how close this may be to SHR. 

 

The work stuff you asked about does not seem to change. Samsung is one of the companies I work with a great deal. Ironically an IT communications company based in Korea. Assigned to a large retail account I am tasked with maintaining digital signage within retail stores by a given brand for USA and Canada. Upon Covid many retailers used digital signage to support reductions in staff counts. Guess that makes it automation in a sense. In the past I have also worked with Humax. Depending on size of job overnight shifts become mandatory in retail. Arrays or video walls are turning in to frequent service items. Client likes to put heat trapping shrouds around the arrays hiding the mechanicals thus making it impossible for BTU load to exhaust. Invites frequent visits with scissor lift coordination and overnight work. Looking back at my posts, yes....sleep is important. 

 

@Vodka2014 have you tried rebooting with only one of the drives connected? Want to say that DSM installs to each drive. If it was me I would alternate drives on reboots looking for one that works. If you boot successfully and see interface inserting remaining drives at that point may cause repair link to show in storage manager. 

  • Like 2
Link to comment
Share on other sites

I tried tinycore-redpill.v0.9.4.0-1.m-shell this morning however I am not booting successfully. 

 

/root/boot/.sh: line 588: [: 0: unary operator expected

 

later the actual files error is:

 

Cannot open /mnt/tcrp/zImage-dsm: no such file or directory. It also looks like this tried to boot as non-efi on a guid partition map, may be the larger indicator

Link to comment
Share on other sites

Made another usb stick and got past error. Using 3622XS+. 2 NVME drives on motherboard are now getting recognized. On friend screen notice that there are two commands reported as not found:

 

DiskIdxMap: command not found

SataPortMap: command not found 

 

First line of boot arguments reads: withefi. This was with v9.3.0-m-shell image. The NVME's show as detected but do not show up on any volume as SSD cache. When I hit create and select SSD cache it says no drives available. Right clicking on an nvme and selecting erase shows indication that it's associated to a storage pool and I would need to remove it first. Storage manager appears to be locked on the NVME's. Can't change any states. NVME's are not attached to any existing pool. I unchecked the box on each for enable SSD cache and rebooted to see no change. Is there a trick to freeing up the NVME's so they don't appear associated to a pool reflecting accurate state?

Link to comment
Share on other sites

23 hours ago, daryl said:

Made another usb stick and got past error. Using 3622XS+. 2 NVME drives on motherboard are now getting recognized. On friend screen notice that there are two commands reported as not found:

 

DiskIdxMap: command not found

SataPortMap: command not found 

 

First line of boot arguments reads: withefi. This was with v9.3.0-m-shell image. The NVME's show as detected but do not show up on any volume as SSD cache. When I hit create and select SSD cache it says no drives available. Right clicking on an nvme and selecting erase shows indication that it's associated to a storage pool and I would need to remove it first. Storage manager appears to be locked on the NVME's. Can't change any states. NVME's are not attached to any existing pool. I unchecked the box on each for enable SSD cache and rebooted to see no change. Is there a trick to freeing up the NVME's so they don't appear associated to a pool reflecting accurate state?

 

The message below of TCRP FRIEND is not exactly an error situation.

DiskIdxMap: command not found
SataPortMap: command not found

For the automatic mapping of SataPort, the part that was deliberately processed in the M SHELL
TCRP FRIEND just detects it as an error.

 

The two NVMe caches are recognized, but I don't understand why You can't add caches to volumes in a storage pool.
Please upload at least one screenshot of the current screen of your disk manager.
I think it will be quicker to understand what you're explaining by looking at the disk manager.

Edited by Peter Suh
Link to comment
Share on other sites

Thought to mention the errors. Glad they are not critical. The NVME's currently show as part of SSD Cache Group2 but when I look at the volumes that group is not showing up on any as associated. If I try to create a group it says that no available SSD's are available for. I may be new at assigning SSD cache but I can't figure out how to remove SSD Cache Group 2 from the nvme's so that they can show as available to storage manager. Please forgive if this is a stupid question and thank you. Seeing some old files when I look around. I've changed DSM versions many times. Saw a DS1019+ file and have been deleting them as I find however I can't help but think that I may have an old config file that may be locking these to a different model....? I see synonvme command. Is it possible to find owner with?

 

1071171327_Screenshot2023-01-29at12_49_57PM.thumb.png.3cf66308c80f581315c27bb65089797a.png1839678386_Screenshot2023-01-29at12_50_44PM.thumb.png.b6fd8cf54ae989d933a1396e01a93cd4.png816311196_Screenshot2023-01-29at12_50_13PM.thumb.png.3c414b52afeb7d1af159981bbd69ccc4.png

Edited by daryl
updated to mention many DSM installs so far and old config file possibility
Link to comment
Share on other sites

7 hours ago, daryl said:

Thought to mention the errors. Glad they are not critical. The NVME's currently show as part of SSD Cache Group2 but when I look at the volumes that group is not showing up on any as associated. If I try to create a group it says that no available SSD's are available for. I may be new at assigning SSD cache but I can't figure out how to remove SSD Cache Group 2 from the nvme's so that they can show as available to storage manager. Please forgive if this is a stupid question and thank you. Seeing some old files when I look around. I've changed DSM versions many times. Saw a DS1019+ file and have been deleting them as I find however I can't help but think that I may have an old config file that may be locking these to a different model....? I see synonvme command. Is it possible to find owner with?

 

1071171327_Screenshot2023-01-29at12_49_57PM.thumb.png.3cf66308c80f581315c27bb65089797a.png1839678386_Screenshot2023-01-29at12_50_44PM.thumb.png.b6fd8cf54ae989d933a1396e01a93cd4.png816311196_Screenshot2023-01-29at12_50_13PM.thumb.png.3c414b52afeb7d1af159981bbd69ccc4.png

 

Your SSD Cache Group 2 looks like a ghost, not assigned to any volume. Is that correct?
I think there will be an answer on the Synology KB,
Under normal circumstances it should look something like this:

 

1846671406_2023-01-3010_11_38.thumb.png.63d04618af2fdb8d092b2ff2342d1ec9.png

 

Is it possible to select cache on this screen and process "Deactive Drive" in Action menu?

 

 

Edited by Peter Suh
Link to comment
Share on other sites

I will search the synology kb. The options are basic. Benchmark, erase and configure write cache. Reset says it can't. Configure write cache is the the pop up that allows one to check box for include or not. Neither states change anything. I have a feeling an old config file is locking the permissions on them. 

 

I was poking around in dev rules and found a 10-unlock-disks.rules file and copied out

 

/usr/syno/sbin/synostorage --unlock-disk

 

aimed it at nvme0 and nvme1 with sudo. Rebooted and no change. Check boxes are off enable write cache pop up  

 

sudo synostorage --lock -d /dev/nvme0 and again for nvme1  to no avail either

 

Saw something like this once when an apple mouse wouldn't recognize on the computer until the thing was reboot without a mouse plugged in and once mouse connected it recognized. Its late here. I will take the nvme's out tomorrow understanding that if they actually are associated I may loose data. If not booting without them in and then rebooting again with them inserted may add to my file lock suspicion should they not be freed up. I will post results tomorrow.

 

Yes you are correct not associated and a ghost. Previous models installed were DS1621+ ryzen and the DS1019+. The DS1621 used both in bidirectional and I mounted to an SSD. Worked well. The DS1019 only used one as you know. I may pull all drives but the 120g if the above method does not work and install new and see how it looks on reboot. Hey, it seems like old configs are backed up on the usb. These go in an .xzm and not readable to the system on boot? 

Edited by daryl
Added method 2 and question on xxm
Link to comment
Share on other sites

33 minutes ago, daryl said:

I will search the synology kb. The options are basic. Benchmark, erase and configure write cache. Reset says it can't. Configure write cache is the the pop up that allows one to check box for include or not. Neither states change anything. I have a feeling an old config file is locking the permissions on them. 

 

I was poking around in dev rules and found a 10-unlock-disks.rules file and copied out

 

/usr/syno/sbin/synostorage --unlock-disk

 

aimed it at nvme0 and nvme1 with sudo. Rebooted and no change. Check boxes are off enable write cache pop up  

 

sudo synostorage --lock -d /dev/nvme0 and again for nvme1  to no avail either

 

Saw something like this once when an apple mouse wouldn't recognize on the computer until the thing was reboot without a mouse plugged in and once mouse connected it recognized. Its late here. I will take the nvme's out tomorrow understanding that if they actually are associated I may loose data. If not booting without them in and then rebooting again with them inserted may add to my file lock suspicion should they not be freed up. I will post results tomorrow.

 

Yes you are correct not associated and a ghost. Previous models installed were DS1621+ ryzen and the DS1019+. The DS1621 used both in bidirectional and I mounted to an SSD. Worked well. The DS1019 only used one as you know. I may pull all drives but the 120g if the above method does not work and install new and see how it looks on reboot. Hey, it seems like old configs are backed up on the usb. These go in an .xzm and not readable to the system on boot? 

 

You probably changed your Synology model in the following order regarding NVMe cache.
DS1621+ -> DS1019+ -> DS3622xs+
Each model differs in how it detects NVMe as a cache.


DS1621+ DTC mapping by dtbpatch driver.
DS1019+ is a Hexa Edit of the libsynonvme.so.1 file.
DS3622xs+ write to /etc.defaults/extensionPorts file

 

From the DS1621+, it seems that you have created a dual cache group with read/write on one of the three volumes,
In a situation where only one was seen in DS1019+, it seems that this dual cache group was abnormally twisted.
Then with the DS3622xs+ the situation seemed to get worse.
Is that right?

 

If possible, I will prepare a bare metal that can use dual NVMe and test it according to the scenario above.
Please check again if this scenario is correct.

Link to comment
Share on other sites

15 hours ago, TrickyTrix said:

In part 7.Build Boot Loader After SSH Connection (Two Step Update - TSU) syntax should be changed.
"./rploader.sh build apollolake-7.1.0-42661" is "./rploader.sh build ds918p-7.1.1-42962" now.

 

Unfortunately, once a post has been created, it cannot be edited after a certain period of time.
I already know about it, but I can't update it.
thank you

Link to comment
Share on other sites

For anyone that does what I did, that Peter correctly labeled:

 

DS1621+ -> DS1019+ -> DS3622xs+

 

I found that I could get control of my NVME's back by booting to TCRP and using fdisk to delete the partitions on nvme0 and nvme1. I'm willing to bet they were still attached to the storage pool previously assigned as I am now scrubbing data and had no choice. Pool in warning status after doing the partition deletion and rebooting but at least they are showing up in the gui now, can be removed, reassigned. So forth. 

 

Peter, don't go to that much work. Willing to bet a new partition layout which is mush easier to script with say sgdisk would be easier to throw at the NVME drives should this seem to be user repeated.

 

I actually went to DS1621XS+ seeing a new possibility and at first on this model I had no NVME's showing. I did what I had mentioned and removed NVME's and restarted TCRP. Reinstalled NVME's and it really didn't do anything except trigger a data scrubbing flag (BTRFS). This was key as I then went to storage manager and started data scrubbing and immediately paused and reboot. On reboot I was showing one nvme so I got the DS3622 target in mind again but deleted both NVME partitions before installing it and viola. Mandatory scrubbing once done but I am fully configurable on NVME once again. 

 

@Vodka2014 that is really good news, glad it worked

Edited by daryl
Full explanation on path of realization
  • Like 1
Link to comment
Share on other sites

12 hours ago, daryl said:

For anyone that does what I did, that Peter correctly labeled:

 

DS1621+ -> DS1019+ -> DS3622xs+

 

I found that I could get control of my NVME's back by booting to TCRP and using fdisk to delete the partitions on nvme0 and nvme1. I'm willing to bet they were still attached to the storage pool previously assigned as I am now scrubbing data and had no choice. Pool in warning status after doing the partition deletion and rebooting but at least they are showing up in the gui now, can be removed, reassigned. So forth. 

 

Peter, don't go to that much work. Willing to bet a new partition layout which is mush easier to script with say sgdisk would be easier to throw at the NVME drives should this seem to be user repeated.

 

I actually went to DS1621XS+ seeing a new possibility and at first on this model I had no NVME's showing. I did what I had mentioned and removed NVME's and restarted TCRP. Reinstalled NVME's and it really didn't do anything except trigger a data scrubbing flag (BTRFS). This was key as I then went to storage manager and started data scrubbing and immediately paused and reboot. On reboot I was showing one nvme so I got the DS3622 target in mind again but deleted both NVME partitions before installing it and viola. Mandatory scrubbing once done but I am fully configurable on NVME once again. 

 

@Vodka2014 that is really good news, glad it worked

 

I'm glad that you can use NVMe normally.
The disk initialization and commands in TCRP, which I like to use, are as follows.
I've been using it mainly for disks, but I think it can be applied to NVMe as well.

 

sudo dd if=/dev/zero of=/dev/sda bs=1k count=1;sudo blockdev --rereadpt /dev/sda;

Link to comment
Share on other sites

9 hours ago, rufik said:

I'm running 7.1-42661 on loader v0.8 and I'd like to upgrade to latest 7.1.1-429xx and tcrp v0.9.4. So quick question - should I go with full guide (all steps) or only somehow "upgrade" from existing USB pendrive?

 

Going back to tinycore linux from the old loader, you need to update the loader.
The core content of v0.9.4.0 is DS923+ support.

 

Download M SHELL for TCRP below and try it.

curl -L "https://github.com/PeterSuh-Q3/tinycore-redpill/raw/main/my.sh.gz" -O

tar -zxvf my.sh.gz

./menu.sh

 

In the same way, up to DS923+ is supported, and an easy-to-use menu method is applied like ARPL.

Edited by Peter Suh
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...