taos2.2.2.10 报错 Database memory is full，无法建立新表，以及两个 vnode 同步出现问题

[TDengine usage environment] Production environment / test / Poc / pre-production environment [TDengine version] 2.2.2.10 [Operating system and version] CentOS Linux release 7.9.2009 (Core) x86_64 GNU/Linux [Deployment method] Container deployment [Cluster node number] 1 master 2 slaves [Cluster copy number] 3 [Description of business impact] The taos2.2.2.10 version, Linux cluster, one master and two slaves, the mine-arrange library has a problem. There were originally 8 vnodes, and I wrote data into them in batches. I added several hundred tables, and the first 6 vnodes were full of tables, and the last 2 vnodes, vnode75 had 685 tables, and vnode76 had 89 tables. Later, I could not create new tables, saying that memory was full (in fact, it was not), but another host3 slave synchronization always showed synchronization in progress. Later, I restarted it and it still did not work, and it still showed synchronization in progress, but the network speed should be fast. Later, I reduced the cluster of this library to 2, then to 1, and then increased to 3. At this time, the two vnodes no longer had a master (the database could not be queried normally), and when reduced to 2, it was available (queries could be made), and when reduced to 1, there was no problem, but tables could not be written into. What is the reason for this, and what should I do?