hypertable管理当前rangserver和历史rangserver

原创于 2017-03-28 15:11:29 发布 · 441 阅读

0 ·

本内容遵循CC 4.0 BY-SA版权协议

收录于

ht4w 专栏收录该内容

7 篇文章

订阅专栏

本文介绍了一个名为OperationRecoveryBlocker的组件实现细节。该组件用于检查集群中活跃节点的数量是否达到预设的容错比例，如果未达到则阻止进一步的操作执行。文章通过代码示例展示了如何计算当前连接的服务器数量与所需最小存活节点数之间的关系。

purge_old_log_files

void OperationRecoveryBlocker::execute() {

HT_INFOF("Entering RecoveryBlocker-%lld threadid %d", (Lld)header.id, GetCurrentThreadId());

size_t total_servers = m_context->rsc_manager->server_count();
size_t connected_servers = m_context->available_server_count();
size_t failover_pct
= m_context->props->get_i32("Hypertable.Failover.Quorum.Percentage");
size_t quorum = ((total_servers * failover_pct) + 99) / 100;

HT_INFOF("total_servers=%d connected_servers=%d quorum=%d",
(int)total_servers, (int)connected_servers, (int)quorum);

if (connected_servers < quorum || connected_servers == 0) {
block();
HT_INFOF("Only %d servers ready, total servers=%d quorum=%d,"
" waiting for servers ...", (int)connected_servers,
(int)total_servers, (int)quorum);
HT_INFOF("Leaving RecoveryBlocker-%lld state=%s-BLOCKED", (Lld)header.id,
OperationState::get_text(get_state()));