UEFI installation via httpboot issue

General support questions
Post Reply
snipergr
Posts: 4
Joined: 2020/10/09 16:25:10

UEFI installation via httpboot issue

Post by snipergr » 2020/10/09 16:47:32

Hello,

I decided to give a shot to httpboot instead of the traditional PXE boot.
Everything works as expected, grub.cfg (after all needed files were fetched and downloaded from the HTTP server) is being loaded and the menu to select the OS to be installed appears.
After selecting the appropriate entry (only one - Centos7-1908-minimal), I can see that vmlinuz and init ramdisk are being requested and server responds successfully with status 200. Despite this, after selecting the entry in order to load the kernel and the ramdisk and to initiate the installation procedure, I am having the well known:
No filesystem could mount root
Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)
Of course, this is followed by a call trace, and system hangs.

Grub.cfg file:
---
set default 0
set timeout=10

menuentry 'Install Centos 7 Minimal' {
echo "Loading vmlinuz"
linuxefi /correct/path/to/vmlinuz ip=dhcp inst.repo=http://x.x.x.x/centos7-1908-minimal/
echo "Loading initrd.img"
initrdefi /correct/path/to/initrd.img
}
---
This is the same file that could be used on a TFTP-based PXE installation.

Am I missing something here? Maybe inst.stage2 parameter?

Regards,
Chris

User avatar
TrevorH
Forum Moderator
Posts: 29681
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: UEFI installation via httpboot issue

Post by TrevorH » 2020/10/09 17:10:48

That means there was a problem either fetching the initramfs or unpacking it and mounting it. Or it's corrupted. Possibly all of those. It's also possible that you don't have enough RAM since all these things are unpacked and mounted as 'ram disks' so lack of RAM can lead to filesystem truncation and corruption.
CentOS 6 will die in November 2020 - migrate sooner rather than later!
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 is dead, do not use it.
Full time Geek, part time moderator. Use the FAQ Luke

snipergr
Posts: 4
Joined: 2020/10/09 16:25:10

Re: UEFI installation via httpboot issue

Post by snipergr » 2020/10/09 17:28:03

Hey Trevor,
- RAM is certainly not an issue, so we can exclude that.
- Regarding corruption, it is the same initrd.img extracted from the ISO image (located at /images/pxeboot/ folder).
- Regarding the three other possible problematic points, fetching, unpacking and mounting. The truth is that I just gathered a pcap file on http server, and I observed this behavior: Target system is requesting for the initrd.img file from the http server (issuing a HEAD and afterwards a GET request for this file), and after 1 or 2 seconds, http server packets stop being acknowledged by the target system. And of course, no initrd.img file seems to be sent. So, I guess, that initrd.img is never downloaded? (despite the 200 response from the HTTP server). Is there any way to further troubleshoot this situation?

User avatar
TrevorH
Forum Moderator
Posts: 29681
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: UEFI installation via httpboot issue

Post by TrevorH » 2020/10/09 17:34:47

Be specific: how much RAM?
CentOS 6 will die in November 2020 - migrate sooner rather than later!
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 is dead, do not use it.
Full time Geek, part time moderator. Use the FAQ Luke

snipergr
Posts: 4
Joined: 2020/10/09 16:25:10

Re: UEFI installation via httpboot issue

Post by snipergr » 2020/10/09 17:37:10

192G

User avatar
TrevorH
Forum Moderator
Posts: 29681
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: UEFI installation via httpboot issue

Post by TrevorH » 2020/10/09 18:01:39

And is it all being recognised by the system?
CentOS 6 will die in November 2020 - migrate sooner rather than later!
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 is dead, do not use it.
Full time Geek, part time moderator. Use the FAQ Luke

snipergr
Posts: 4
Joined: 2020/10/09 16:25:10

Re: UEFI installation via httpboot issue

Post by snipergr » 2020/10/12 17:51:55

Yes.
Today I have also tried mem and ramdisk_size parameter on grub.cfg, but no luck.
Took another pcap file on server acting as HTTP server, these are the last packets received (this occurs in less than 1 second):
- After receiving vmlinuz, target system sends a HEAD request for initrd.img file
- HTTP server responds with 200 0K, with content-length info.
- Target system issues a GET request for initrd.img file.
- HTTP server responds with an ACK and a PSH,ACK.
- After that, HTTP server tries to communicate with target system (continuous ACKs- same SEQ number), but no response.

Unfortunately, as I am viewing the console remotely, I cannot scroll up in order to check previous logs. Is there any way to redirect those logs to a remote rsyslog server? Maybe there is a helpful error log on what is happening.

Post Reply

Return to “CentOS 7 - General Support”