CA-412636: hostname changed to localhost with static IP and reboot

[liulinC]

According to
https://www.freedesktop.org/software/systemd/man/latest/hostname.html Systemd set hostname with following sequence

• kernel parameter, systemd.hostname
• static hostname in /etc/hostname
• transient hostname like DHCP
• localhost at systemd compile time

Once the host is configured with static IP and reboot, it would just lost its hostname as no DHCP or static IP available.

However, the hostname is critical to AD function as it construct the machine account.

The hostname should be persisted as static name during joining AD, this is also what PBIS does.

Note: the static hostname is not cleaned during domain leave. This is by intention to avoid losing hostname after reboot with static IP

cba2f1d tried to resovle the issue and update /etc/resolv.conf However, /etc/resolv.conf does not help and conflict with xcp-networkd, as networkd override the configure every 5 minutes Here we just revert the resolv.conf update

[edwintorok]

I can't find commit 0307adc6, do you have the full git hash and title?

[liulinC]

I can't find commit 0307adc6, do you have the full git hash and title?

Sorry, I got the commit ID from wrong branch (my dev branch before merged to master), thus wrong commit id.
it should be cba2f1d, and updated the commit message.

[lindig]

So the approach of rewriting resolve.conf was wrong? But this was done for a purpose - it that still achieved?

[liulinC]

So the approach of rewriting resolve.conf was wrong? But this was done for a purpose - it that still achieved?

writing resolve.conf is NOT wrong, on the contrary, it is right
It absolutely make sense to write domain into resolv.conf . (there is a customer do expect that and there is an XSI for that).

However, simply write to that file conflict with networkd, I may update later to call networkd to update the domain later, (update_config). but that will require more test.

This PR overriding the hostname just follows what PBIS does, and it resolve the localhost issue.

[edwintorok]

Aren't we going to regress the fix for XSI-1918 though? I'd rather wait until we have a fix that works for both CA-412636 and XSI-1918 and merge it in the same PR so we don't risk breaking AD for existing users.

[lindig]

I am also confused about the relationship between the code that is removed and the new code - hence my question earlier. This PR should present more context how these problems are related.

[liulinC]

1918

The deleted code does not fix XSI-1918.
The real fix for XSI-1918 is:

• update hostname order in /etc/hosts. e.g: the un-removed part of cba2f1d
• This PR, e.g: set static hostname during join pool.

The removed code does not have any issue as it was override by xcp-network anyway. (and It will not cause customer any trouble)

[liulinC]

I am also confused about the relationship between the code that is removed and the new code - hence my question earlier. This PR should present more context how these problems are related.
in XSI-1918, we found two main issues.

1. The hostname is no write into /etc/hosts in right order, as stated in cba2f1d, and in that PR we

• fix the order issue
• push hostname to DNS server by best effort. (not a regression), so pool numbers can resolve each other by hostname with static IP setting
• write domain to /etc/resolv.conf. so the default domain is the join domain. (not a regression, but an enhancement)

2. hostname should be set to static hostname like PBIS during pool Jon.

• This PR fix the issue.
• I found that the /etc/resolv.conf configuration is conflict with scp-networkd. and it is an enhancement, (Not the issue reported by XSI-1918). So I revert the change in this commit.

[edwintorok]

I see, so this only partially reverts the previous change, in particular the resolv.conf update which introduced a bug as it conflicted with networkd, and in addition it sets the hostname using an external command.