Fresh install of version 2.8 fails to boot
-
You'd have to contact the manufacturer to get a BIOS update and cross your fingers!
But that may not help if it's installing legacy and booting uefi. It probably has a BIOS option to boot legacy installs. But you should be able to install uefi.
The installer itself can boot both legacy or UEFI and will default to installing the same mode as it booted. But some systems will boot USB legacy and SSDs UEFI only. You may simply be seeing that. Check what options it's installing with.
-
@stephenw10
I'm backing off my earlier analysis a bit.I just reinstalled the OPNsense drive and it,, again, works great.
I think the installer has an issue that's far beyond my skill level to even understand. I don't think it writes the boot tracks correctly sometimes.
I have just pushed it around a lot and it still works. I ran an update to a newer version, some misc DhCP reservations on 2 subnets. I updated adguard home. No problems. It worked as expected.
The Chinese Box has a 9-2024 bios date. 4 i226v ports. J4125 CPU. It's new.
What could be different between what was written to that ssd and the pfSense ssd during pfSense installs? And it worked great up until the 2.8 - 2.8.1 upgrade that hosed up the plugin installer.
My main router runs pfSense well. It's a different Chinese box. It's on 2.8 CE. The updater has always been unable t find 2.8.1. Now I am happy it has because my fiasco started with a 2,8 - 2.8.1 upgrade. But now I am unsure about future reliability.
I prefer pfSense because I know my way around it and it has a couple of minor features that are actually important to me that OPNsense does not have. Please look into the installer for subtle bugs or incompatibilities in how the boot tracks are written. The pfSense boot has some loopiness in it - I can make it work if I go into the bios and tweak the boot, but the tweak won't stay permanent. OPNsense and old pfSense on it could boot and boot and boot all day.
-
Hi @stephenw10 ,
i did a fresh installation of pf sense CE 2.8.1 on my microserver gen 10 plus, and this the last image i see before it restarts.
-
@coffeecup25 said in Fresh install of version 2.8 fails to boot:
What could be different between what was written to that ssd and the pfSense ssd during pfSense installs?
Was it installed on the same hardware?
The most important thing is what settings you selected during the install. Specifically did you set MBR or GPT?
https://docs.netgate.com/pfsense/en/latest/install/install-walkthrough.html#filesystem-and-partition-settingsIt's almost certainly not booting because it's trying to boot UEFI and there isn't a bootable partition present. It should show an error telling you that at boot.
-
@stephenw10 said in Fresh install of version 2.8 fails to boot:
@coffeecup25 said in Fresh install of version 2.8 fails to boot:
What could be different between what was written to that ssd and the pfSense ssd during pfSense installs?
Was it installed on the same hardware?
The most important thing is what settings you selected during the install. Specifically did you set MBR or GPT?
https://docs.netgate.com/pfsense/en/latest/install/install-walkthrough.html#filesystem-and-partition-settingsIt's almost certainly not booting because it's trying to boot UEFI and there isn't a bootable partition present. It should show an error telling you that at boot.
Same hardware. It first failed after the 2.8 - 2.8.1 upgrade where I got the message that the plugins could not be upgraded because another upgrade was in process. Even well after the upgrade had 'successfully completed' I read on the internet this message is a 'forever message' if you can't get rid of it in a normal time period. I found some commands that were supposed to stop the upgrade and entered them. They didn't work. pfSense did not reboot after one.
I decided to go for a clean install and downloaded a new usb drive. It had the ability to load 2.8.1, 2.8, 2.7.2 I used Rufus default settings to make the usb.
The USB was recognized after I changed the boot order. It booted every time I tried the re-install. Probably 10 times.
All end product loads failed to boot, although I could coax the last couple of tries by - and this is weird - trying and failing, then going into the bios and seeing the crucial drive and setting it to the top, exit and reboot. Then success. BUT a reboot failed because the set didn't work.
I put my OPNsense drive back in and it fired up and I did my normal upgrades there, the ones I had planned to do to this box with pfSense - which is my spare router and lab.
I also took out the drive and tested it on my pc with a usb attachment. I wiped it and reformatted it to ntfs using a partition manager. It worked as it should. I have seen bad drives, and this one worked as it should. If anything stood out I would have replaced it with another spare ( which I may do anyway later just to see if it works.)
The mfgr recommended I try with something other than pfSense to see if it works. I plan to try that in the next couple of days. They agreed it sounds like an installer problem. I think a bug got in there somehow and bad boot tracks are written.
The Chinese Box has an American Megatrends bios dated 9-2024. The name of the company is HKUXZR. From Amazon - Micro Firewall Appliance, 4 Inter i226-V 2.5Gbe LAN Ports, Fanless Mini PC Celeron J4125 Quad Core, DP/HDMI USB3.0 Support AES-NI Barebone Router PC VPN (NO RAM NO SSD)
If everything else is found to work but pfSense then I will be forced to move to OPNsense for the sake of reliability. This is not my first choice as I really do not want to learn a new interface. I also have a TP-Link ER605 V2 that is surprisingly capable but not as much fun.
-
@coffeecup25 said in Fresh install of version 2.8 fails to boot:
I could coax the last couple of tries by - and this is weird - trying and failing, then going into the bios and seeing the crucial drive and setting it to the top, exit and reboot.
Ok that seems to add to my suspicion that it's just not trying to boot the installed partition. If you manually select it then it does boot.
So you need to check what your BIOS is trying to boot.
It might be trying to boot legacy only or UEFI only and the installed partition is the other type. Or it might be not have an EFI variable value for the new partition.
We need more info to know though. Like what entries doe the BIOS show and what does it label those as. Or what does it show when fails to boot?
-
This post is deleted! -
@stephenw10 said in Fresh install of version 2.8 fails to boot:
@coffeecup25 said in Fresh install of version 2.8 fails to boot:
I could coax the last couple of tries by - and this is weird - trying and failing, then going into the bios and seeing the crucial drive and setting it to the top, exit and reboot.
Ok that seems to add to my suspicion that it's just not trying to boot the installed partition. If you manually select it then it does boot.
So you need to check what your BIOS is trying to boot.
It might be trying to boot legacy only or UEFI only and the installed partition is the other type. Or it might be not have an EFI variable value for the new partition.
We need more info to know though. Like what entries doe the BIOS show and what does it label those as. Or what does it show when fails to boot?
when it fails it only shows some default setting. no drives are listed.
With the usb installer, it shows the usb to select and some other generic info
After install and after 1st fail, it shows the ssd to select and boots properly.
on reboot it goes back to the default with no drives.
I think I can't detect boot tracks properly on the pfSense ssd. That would make it an installer problem. pfSense was installed using defaults with WAN and LAN selected. My phone is WAN with usb connector and PC is LAN.
If there is a user problem, then it is too obscure to keep working on at my end. And you will see it more from others as time passes.
-
Is it set to boot UEFI only or legacy and EFI?
Can you get a screenshot?
-
@stephenw10 It only has one setting. I think it's uefi as I think that is the current modern default for a lot of hardware.
No screen shot as it's put away after a over the top yesterday with it. If it is the user, then it's still your problem with the installer as it is a new pc as in new design and I'm using defaults substantially. The mfgr sells a lot of hardware. American Megatrends is not ACME. If it were old junk, As I wrote earlier I would switch from legacy to uefi or vice versa as that problem is old news to me. I don't think it has a legacy mode, but I'm only guessing as I did not see one in the obvious places to check.
Also, it worked 100% perfectly well a few months ago when I loaded pfSense on it the 1st time (2.8) The problems occurred with the current installer after i totalled out the original pfSense install trying to fix a different bug.
-
Hmm, odd. Most of those Chinese boxes use a pretty generic BIOS with a lot of exposed options.
You never said whether you installed MBR or GPT? Or what the installer selects there by default.
Another thing you can check is boot the installer then drop to the rescue shell and run:
sysctl machdep.bootmethod
That will show you how it's booting when it boots from USB.You could also try that from any booted SSD, including OPN. If it's booting differently there that's almost certainly the problem.
-
@stephenw10 said in Fresh install of version 2.8 fails to boot:
Hmm, odd. Most of those Chinese boxes use a pretty generic BIOS with a lot of exposed options.
You never said whether you installed MBR or GPT? Or what the installer selects there by default.
Another thing you can check is boot the installer then drop to the rescue shell and run:
sysctl machdep.bootmethod
That will show you how it's booting when it boots from USB.You could also try that from any booted SSD, including OPN. If it's booting differently there that's almost certainly the problem.
As I mentioned no MBR, legacy, options to select that I could see. AND it worked perfectly before a few months ago and I guaranty I did not play around with bios settings.
I installed whatever your default is on the installer. Like always.
I'll try the other thing in a couple of days.
I completely understand the mbr vs gpt choices. As I said, the loader usb stick always booted perfectly after being built by Rufus using defaults. I used YOUR INSTALLER defaults now and a few months ago when it worked perfectly. With that information you should be able to back into the answer for your question. I'm not going to play gotcha.
I plan to load OPN in a day or two on the same ssd to see if they have a boot problem. If no problem, then it's the pfSense installer. If there's a problem, I'll try another ssd. If there's still a problem then it's the PC. If not a problem with the new ssd, I will again try 2.8.1. If still a problem then it's the installer and how it writes boot tracks.
Then fix the 2.8 error that hides the 2.8.1 available upgrade. Then fix the occasional endless plugin installer problem that happens after the 2.8.1 upgrade.
-
As I said the installer itself with boot either. But it uses the detected boot type to select the default filesystem for the install. That is an option in the installer, it's always present. You may need to set it manually if your hardware/BIOS boots USB differently to the SSD.
As as fallback you can always install 2.7.2 from the legacy installer and then upgrade.
-
@stephenw10 said in Fresh install of version 2.8 fails to boot:
As I said the installer itself with boot either. But it uses the detected boot type to select the default filesystem for the install. That is an option in the installer, it's always present. You may need to set it manually if your hardware/BIOS boots USB differently to the SSD.
As as fallback you can always install 2.7.2 from the legacy installer and then upgrade.
I will try what I said using OPN. If it works I will learn the new interface. pfSense CE is slowly becoming unstable. See the other 2 bugs for reference. they are small but the length of time they have existed makes me think CE is not a priority.
I used to program long ago and I was an ace debugger of other people's work along with being really good at original projects. The first debug issue was to nail down where it likely exists. I actually stopped to figure out the actual problem rather than run a debugger and trace through it until it ran well enough to walk away from.
MBR vs GPT is not the issue, and if it is then it's still your problem because it never has been a problem before, even with your older installers or with other software. If it is the problem then it's still your problem for apparent lack of interest on making it easy to deal with. Believe it or not, very few users want to make it a serious project to make software work if substitutes are available that don't have similar problems. I'm not going to go to 2.7.2 and upgrade from there. Ridiculous. I would take the ER605 v2 out of storage first. (BTW, I wrote TP-link about a dhcp DNS issue ,they had only 2 selections available for overrides like almost everyone else and I wondered if they could be like pfSense and make more available. I just looked today and down at the device assignment level, you apparently can have a list of servers listed. Unexpected.)
-
If there's some bug here I'd love to squash it but I've got nothing to work with so far. I can only guess why it's not booting for you without logs or screenshots.
The closest to what you've described I can replicate is trying to boot using an incompatible method but, I agree, I have no idea why that would have changed in a few months.
-
@stephenw10 said in Fresh install of version 2.8 fails to boot:
If there's some bug here I'd love to squash it but I've got nothing to work with so far. I can only guess why it's not booting for you without logs or screenshots.
The closest to what you've described I can replicate is trying to boot using an incompatible method but, I agree, I have no idea why that would have changed in a few months.
Then hand the problem over to someone with a more active imagination. I have written a book here telling you about it. And I suspect the issue would devolve into a gotcha game, like now only more off point.
I'll try to fix it. If I can't then you can feel good about yourself because I didn't say the magic word needed to get you people moving.
-
I imagine you're having a bad day and I'm sorry about that.
But I have no way to try to replicate the issue you're seeing.
-
@stephenw10 said in Fresh install of version 2.8 fails to boot:
I imagine you're having a bad day and I'm sorry about that.
But I have no way to try to replicate the issue you're seeing.
Thanks for all your hard work on this. I know how hard you tried.
I reinstalled the pfSense ssd back into the offending Chinese box. It continued to fail to boot reliably. Weirdly, it booted into pfSense now every other time or so - no consistent fail or success.
Then I loaded OPNsense from the same thumb drive that earlier held the pfSense installer as a trial to see if it worked perfectly like the OPNsense ssd I just took out.
Then, after it loaded and ran perfectly I rebooted it several times to see if it stayed on point and didn't flake out into that weird shell.
It just keeps running.
And I will keep running it as pfSense CE has become unreliable. I'm using the new setup now to write this and my cell phone via USB adapter is the WAN. In a couple minutes I will reload a backup configuration and then relearn how to side load Adguard Home and reconfigure than from backup as well.
Thanks for all the really great (and I mean it) years of service up to this point. My main router is a different Chinese box and it still runs pfSense with AdGuard Home running in the background. I do not plan to swap it out just yet. But I feel secure having 2 backup routers 1) The OPNsense one I am building now but still have a fully configured ssd in reserve. And 2) an extremely capable TP-Link ER605 V2 combined with a pair of currently running servers holding AdGuard Home.
slight correction: I hadn't switched over the LAN cable to write the above. NOW I am writing from the new config.
-
Hmm. Inconsistent booting from the same device without changing anything is even more weird. Hard to imagine what could cause that. Some slight timing variation perhaps. But from an SSD that's unexpected.
