[LoadBalance] Optimize find nics process.

[mattisonchao]

Motivation

According to this PR #14252,

We can know In some cases, VMs in EC2 won't have the speed reported on the NIC and will give a read error.

In a normal case, only eth0 has speed value. other devices have no speed value.

So when we check isPhysicalNic , line 232 would get a read error many times.

pulsar/pulsar-broker/src/main/java/org/apache/pulsar/broker/loadbalance/impl/LinuxBrokerHostUsageImpl.java

Lines 229 to 247 in d3848e2

	private boolean isPhysicalNic(Path path) {
	if (!path.toString().contains("/virtual/")) {
	try {
	Files.readAllBytes(path.resolve("speed"));
	return true;
	} catch (Exception e) {
	// In some cases, VMs in EC2 won't have the speed reported on the NIC and will give a read-error.
	// Check the type to make sure it's ethernet (type "1")
	try {
	String type = new String(Files.readAllBytes(path.resolve("type")), StandardCharsets.UTF_8).trim();
	return Integer.parseInt(type) == 1;
	} catch (IOException ioe) {
	// wireless nics don't report speed, ignore them.
	return false;
	}
	}
	}
	return false;
	}

Then I think we can change the process to check type first and then to check speed.

Modifications

• Change find NICs process.
• Add log to remind the user to set config loadBalancerOverrideBrokerNicSpeedGbps when we can't read speed that is ethernet.

Verifying this change

• Make sure that the change passes the CI checks.

Documentation

• no-need-doc

[mattisonchao]

/pulsarbot rerun-failure-checks

[Technoboy-]

Could you help review this patch, for we're not sure about it.
@merlimat @codelipenghui @eolivelli @lhotari .

[mattisonchao]

Supplementary Instructions:

Linux /sys/class/net/<iface>/type values: https://github.com/torvalds/linux/blob/master/include/uapi/linux/if_arp.h
Linux /sys/class/net/<iface>/speeddefination:

Indicates the interface latest or current speed value. Value is an integer representing the link speed in Mbits/sec.
Note: this attribute is only valid for interfaces that implement the ethtool get_link_ksettings method (mostly Ethernet).

[mattisonchao]

/pulsarbot rerun-failure-checks

[BewareMyPower]

Should it be cherry-picked to other active branches? @codelipenghui

I also found this problem in EC2 VMs.

[michaeljmarshall]

@BewareMyPower - it might be helpful to cherry pick to active branches. In #14537, there was a discussion about failing fast, instead of constantly logging this error message. #14537 (comment)

I expect that behavior will likely get cherry-picked, and that commit might rely on this one.

[Shawyeok]

We had same problem on Xen VM with pulsar 2.8.3, should we pick this to 2.8.x branch? @codelipenghui