Network is halted when updating validator nodes

[eum602]

The problem comes in the following scenario: I have an ibft2 network which is under onchain permissioning.
At the beggining all works well but if for some reason the network stops validating blocks(eg. some validators >1/3 n;n=#total_validators;go offline) then when trying to start the network again by starting the offline validators then the network does not start validating blocks.
The way I found to start the network is by restarting validators without:

--permissions-accounts-contract-enabled
--permissions-accounts-contract-address=0x0000000000000000000000000000000000008888
--permissions-nodes-contract-enabled
--permissions-nodes-contract-address=0x0000000000000000000000000000000000009999

Then the network started without permissioning.

Finally I procceded restarting, gradually in order to not to loose connection between validators. But this time each node was started with the permissioning flags.

If the network fails, then the validators should be able to start again with onchain permissioning without any extra step.

Follow this issue on Hyperledger Besu Rocket Chat

[eum602]

The problem seems to be related with onchain permissioning. Apparently, using bootnodes to help establish communication between validators is possible only if the bootnodes are validators, otherwise if for some reason validators loose communication with other validators and bootnodes are simple regular nodes then validators won't accept communication from other validators and thus the network will remain halted.
https://chat.hyperledger.org/channel/besu?msg=5m5KbNBrusWm2fkiG

[eum602]

Issue has been opened here : https://jira.hyperledger.org/browse/BESU-181