How to safely upgrade (with a non-default-synoinfo)

[IceBoosteR]

Hi everyone,

the newest upgrade from Synology arrived (v.6.2.3) and I do plan do update my box (running at v.6.2.2) to that release.

Now I have 12 disk in my server in two pools (SHR2 - 10 disks) and (RAID1 - 2 disks) so I had to edit the synoinfo file to reflect my storage controller configuration. That was no issue when I started with the box and only applied minor updates afterwards. While the intermediate update is now and intermediatze one, I suggest they will also patch the synoinfo file. So I need to edit this once again after an successful update.

The problem is tho, if I am doing this straight away, my array of 10 disks will not be usable anymore and propably break - as more than 2 disks will become unavailable after reboot. Of course, I want to prevent myself from such a situation. Of course, a backup is available, but honestly, better to not use it,right

 

Some while ago while browsing the forum I found a post from a member who may found a solution for this, but I am unable to find it ...

AFAIK he added a spare drive to his system, before performing the update. Then he disconnects all the drives from his pool (on a physical side eg. pulling of the SATA cable) and reboots with only that one drive. Once that is successful he performs all actions needed and reboots the system with all drives connected again. As the system partition from DSM is now out of sync a quick "repair" is required - but after that all is fine.

So I wanted to ask you how you're doing these updates.

If the above method works, I think I could do it with:

1. Saving the synoinfo file safely

2. Reboot without the SHR2 array connected

3. Perform the update (with only the RAID-1 connected)

4. Edit the synoinfo again

5. Reboot with the SHR2-array connected

6. Rebuild DSM partion

7. Enjoy

 

Any help and/or input is very appreciated

THX

[IceBoosteR]

Anyone? 😮

[flyride]

I would probably do something like that.  Have a backup.

[bearcat]

According to a post from @IG-88  it seems like Synology might have retracted 6.2.3 for older models.

It might be wise to stay calm and wait with the update for now.

 

[IceBoosteR]

5 hours ago, flyride said:

I would probably do something like that.  Have a backup.

Okey. Yes of course backup is available but it is always good to have other thoughts if this is really an appropriate way to do it.

[IceBoosteR]

1 hour ago, bearcat said:

According to a post from @IG-88  it seems like Synology might have retracted 6.2.3 for older models.

It might be wise to stay calm and wait with the update for now.

 

Welp, haven't seen that. Yeah will defenetly wait.

Still, question remains for the future

[IG-88]

dont be shy :-)

 from the change log its just about universal search (you would need to update that plugin to 1.5 before updating to 6.2.3)

i did not have any problems when testing with 6.2.3/918+

tbh my "productive" system still runs on 6.1 and is waiting for it's new hardware to be assembled - but i would use 918+ 6.2.3 if i update - still plan to do this an a daily base ;-)

 

[flyride]

Same here, I keep my archive copy on the latest version but my main system is 6.1.  I am starting the complex system testing to migrate it to 6.2.3, finally...

[IceBoosteR]

20 hours ago, IG-88 said:

dont be shy

 from the change log its just about universal search (you would need to update that plugin to 1.5 before updating to 6.2.3)

i did not have any problems when testing with 6.2.3/918+

tbh my "productive" system still runs on 6.1 and is waiting for it's new hardware to be assembled - but i would use 918+ 6.2.3 if i update - still plan to do this an a daily base

 

  

15 hours ago, flyride said:

Same here, I keep my archive copy on the latest version but my main system is 6.1.  I am starting the complex system testing to migrate it to 6.2.3, finally...

Really, on 6.1?

So I am using the tools and DSM itself heavily and I am really happy with that. As this system is my productive one, I would like to stay at the latest release

Anyhow, so @IG-88 are you using a custom synoinfo with edited port config? This is my main concern, to end up with a failed (SHR-2)-RAID array.

Keep going on your hardware upgrade - today I just received my 24 Bay rackmount case :LOL:
 

Best,

Ice

[IG-88]

3 minutes ago, IceBoosteR said:

are you using a custom synoinfo with edited port config

yes, just max drive 14 (and the other values related in synoinfo.conf)  because my onboard has 4 ports but system/bios blocks 6 ports, so with my 8 port from added controller is 14

 

i had to quickly, as my main unit died (the known atom problem, just 4 years - thanks intel) and i wanted to wait for a 9100T that never came out (yeah i know intel again but that's what synology supports, my desktop is ryzen and as it looks like the next desktop cpu will be again from amd)

 

7 minutes ago, IceBoosteR said:

This is my main concern, to end up with a failed (SHR-2)-RAID array.

i did not test it but the problem might not be that huge when the missing disks are greater then the redundancy, in this case the raid will not start, all raid partitions keep there event number and it should come up again when all disks are back again (or at least as much as needed without the redundancy), the system raid1 is no problem, will fix itself easiely

 

not so good with my config, raid6, positions 13/14 would be missing at 1st boot but raid would start without redundancy disks (event number counting up for 10 of 12 disks) and after fixing maxdisk to 14 i would have to recover the raid with 2 disks, taking days (i can't add a disk for updates as my port count was maxed out, but i made my new system much more flexible)

 

26 minutes ago, IceBoosteR said:

today I just received my 24 Bay rackmount case :LOL:

i'm still planing to change jun's patch o have 24 disks on all 3 extra.lzma's, but not in the next 2 weeks for sure

[IceBoosteR]

1 hour ago, IG-88 said:

yes, just max drive 14 (and the other values related in synoinfo.conf)  because my onboard has 4 ports but system/bios blocks 6 ports, so with my 8 port from added controller is 14

 

i had to quickly, as my main unit died (the known atom problem, just 4 years - thanks intel) and i wanted to wait for a 9100T that never came out (yeah i know intel again but that's what synology supports, my desktop is ryzen and as it looks like the next desktop cpu will be again from amd)

Oh okey thanks.

Where are you living? I've seen some post from you in German, so I assume you do live there. I am using a Supermicro mainboard with a Xeon E3-1270v3 used server gear. It is relatively cheap to purchase, 32GB of ECC memory (which I do use in all my servers) and with 4C/8T more then enough power to run DSM and two VMs. Looking into newer models I have not seen any other CPU-Mainboard combination that could hit that cost-to-power (P/L) ratio. Only looking into the AMD side that would make me to switch, but as you mentioned, Intel only for the DSM...

I can send you a link to the offer where I bought my CPU, offer is still there. Can send you a PM if you like.

 

1 hour ago, IG-88 said:

i did not test it but the problem might not be that huge when the missing disks are greater then the redundancy, in this case the raid will not start, all raid partitions keep there event number and it should come up again when all disks are back again (or at least as much as needed without the redundancy), the system raid1 is no problem, will fix itself easiely

 

not so good with my config, raid6, positions 13/14 would be missing at 1st boot but raid would start without redundancy disks (event number counting up for 10 of 12 disks) and after fixing maxdisk to 14 i would have to recover the raid with 2 disks, taking days (i can't add a disk for updates as my port count was maxed out, but i made my new system much more flexible)

Mhh interesting thought! Unfortunate that it would not work in your config - but yeah, not tested yet so some small risk involved. I guess detaching all disks of that array might be the safest option (including a full backup of course)

1 hour ago, IG-88 said:

i'm still planing to change jun's patch o have 24 disks on all 3 extra.lzma's, but not in the next 2 weeks for sure

Oh okey, cool stuff. Yeah I read that this is now limited, but on the 918+ model only, right? As to my server-grade-hardware, only the 3615xs loader worked for me out of the box (LSI adapter, Mellanox ConnectX-3 10Gbit and the onboard SATA), the others didn't work.

 

I am also looking to upgrade my main rig to AMD, but I spend more money in the past years on server stuff as to my main rig (still i7 6700K+GTX1070) and thats still powerful enough for me

Best,

Ice

[IceBoosteR]

Finally did the upgrade.

Man I swear I will not do this again!

 

I lost the /etc.defaults/synoinfo.conf entirely, which means Synology replaced it with the default. That causes two disks of my SHR-2 to be out of the array, so pool is degraded. I managed to recover everything so disks are back visible (and not esata disks) but the whole procedure is something I do not recommend. Running completely without redundancy is dangerous, and I have seen no other way to recover the file before the system wents online and see, that two disks are missing.

Now the server is shut down as tomorrow electricians want to do some work. I will not risks anything while rebuilding, so I will wait until they finieshed their work. Gladly I did a full backup.

 

Bottom line I did forget about that issue before I did the upgrade, so I also did not pull out the disks before or something. Luckily the update went fine, and now I am hoping the best (fingers crossed). Will do a byte-by-byte-check after the rebuild to make sure everything is fine....

[flyride]

On 4/15/2020 at 11:00 PM, flyride said:

Now, DSM 6.2.3: The upgrade changes at least two fundamental DSM behaviors:

1. SATA devices that are mapped beyond the MaxDisks limit no longer are suppressed, including the loader (appearing as /dev/sdm if DiskIdxMap is set to 0C)
2. The disk port configuration bitmasks are rewritten in synoinfo.conf: internalportcfg, usbportcfg and esataportcfg and on 1.04b, do not match up with default MaxDisks=16 anymore. NOTE: If you have more than 12 disks, it will probably break your array and you will need to edit them back (and that's not just an ESXi issue)!

[flyride]

<deleted>

Edited July 27, 2020 by flyride