fix(meta): refresh metastore if there is zdb up/down

[iwanbk]

part of #127 :

when the meta backend up, zstor monitor automatically reconnect to that backend

[iwanbk]

I got some issue during encryption on

0-stor_v2/zstor/src/encryption.rs

Line 80 in c5f1fba

let plaintext = cipher.decrypt(nonce, &data[AES_GCM_NONCE_SIZE..]).unwrap();

but also happened on the master (before all of my changes), but can't reproduce it again.

I'm checking it now.

[iwanbk]

@LeeSmet

Do you aware of some issue with

0-stor_v2/zstor/src/encryption.rs

Line 80 in c5f1fba

let plaintext = cipher.decrypt(nonce, &data[AES_GCM_NONCE_SIZE..]).unwrap();

?

[iwanbk]

OK, i found the issue, use the wrong config for the encryption

$ git diff
diff --git a/zstor/src/actors/backends.rs b/zstor/src/actors/backends.rs
index 6dd68eb..d9d2929 100644
--- a/zstor/src/actors/backends.rs
+++ b/zstor/src/actors/backends.rs
@@ -400,7 +400,7 @@ impl Handler<CheckBackends> for BackendManagerActor {
                         };
                         let Meta::Zdb(meta_config) = config.meta();
                         let encoder = meta_config.encoder();
-                        let encryptor = match meta_config.encryption() {
+                        let encryptor = match config.encryption() {
                             Encryption::Aes(key) => AesGcm::new(key.clone()),
                         };
                         let new_cluster = ZdbMetaStore::new(

[iwanbk]

i just found the new_metastore func, but it is not used

0-stor_v2/zstor/src/meta.rs

Line 276 in c5f1fba

pub async fn new_metastore(

I tried to use it as well, but it doesn't meet usecase.
If i adjust it, looks like it has to be moved zdb_meta.rs.

If you like, i could do another PR to refactor that code, along with another improvements.
I found that many closures are too long that made it harder to understand

[LeeSmet]

i just found the new_metastore func, but it is not used

0-stor_v2/zstor/src/meta.rs

Line 276 in c5f1fba

pub async fn new_metastore(

I tried to use it as well, but it doesn't meet usecase. If i adjust it, looks like it has to be moved zdb_meta.rs.

If you like, i could do another PR to refactor that code, along with another improvements. I found that many closures are too long that made it harder to understand

Yes, feel free to do so if it is an improvement to the code