[improvement] [client] [naming module] disk cache do not work in some case

[shalk]

The naming module of nacos client have a class service holder.

Service holder have a diskcache feature which is a addition feature for memory cache.

The disk cache feature have some limited , It only take effect when server node changed.

In Production Env, some server nodes is stable and not change for long time.

So I want to improve feature to support diskcache when node is not changed and the file is not exists.

        if (diff.hasDifferent()) {
            NAMING_LOGGER.info("current ips:({}) service: {} -> {}", serviceInfo.ipCount(), serviceKey,
                    JacksonUtils.toJson(serviceInfo.getHosts()));
            
            if (!failoverReactor.isFailoverSwitch(serviceKey)) {
                NotifyCenter.publishEvent(
                        new InstancesChangeEvent(notifierEventScope, serviceInfo.getName(), serviceInfo.getGroupName(),
                                serviceInfo.getClusters(), serviceInfo.getHosts(), diff));
            }
            DiskCache.write(serviceInfo, cacheDir);
        }

Describe the bug
A clear and concise description of what the bug is.

when server node is not change , client disk cache is not enabled

Expected behavior
A clear and concise description of what you expected to happen.
when server node is not change , client disk cache is enabled

Actually behavior
A clear and concise description of what you actually to happen.
when server node is not change , client disk cache is not enabled

How to Reproduce
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See errors

Desktop (please complete the following information):

• OS: [e.g. Centos]
• Version [e.g. nacos-server 1.3.1, nacos-client 1.3.1]
• Module [e.g. naming/config]
• SDK [e.g. original, spring-cloud-alibaba-nacos, dubbo]

Additional context
Add any other context about the problem here.

[KomachiSion]

sorry, i can't get the key point of this issue.

disk cache is used for service application starting when nacos server not worked, let applications start and get the cache from disk to call depend service.

if nacos server recovery, the data should be use server data.

and other feature is a failover cache, the failover is used when nacos server not worked but the depend service ips changed, which need manual change the ip data so that the request not failed.

so I think no need to change this part.

[shalk]

Maybe i make mistake on diskcache generate .

But I found some case the disk cache can not cover ,

Let's focus on disk cache
For example,

consumer A ---RPC--- > producer B ( 3 nodes ,B1, B2, B3,)

1. start nacos server
2. start producer B1,B2,B3, producer B is stable online and registered on nacos for 365 days. not change ips.
3. start consumer A version 1.0 , the disk cache is generated .

---

step1 stop nacos server
step2 restart conusmer A

When consumer A use getAllInstances("serviceName") , the exception always throw.
Is any way to use disk cache when the network request fail.

[shalk]

I think this is bug , it introduted by #7551

because someone found the subscribe is not work in that case. But the fix make the disk cache not work

#7550
#10585

[KomachiSion]

Your comment situation has been fixed by #12929 with duplicate issue #12644