最近链发送交易过大时容易发生一下错误
{
"message": "Internal Service Error",
"code": "500",
"error": {
"code": "3080004",
"name": "tx_cpu_usage_exceeded",
"what": "Transaction exceeded the current CPU usage limit imposed on the transaction",
"details": [{
"message": "transaction was executing for too long",
"file": "transaction_context.cpp",
"line_number": 488,
"method": "checktime"
}, {
"message": "pending console output: ",
"file": "apply_context.cpp",
"line_number": 113,
"method": "exec_one"
}]
}
}
根据错误信息,定位源代码位置 (github)
由于now > _deadline
并且 deadline_exception_code == tx_cpu_usage_exceeded::code_value
走到此断言逻辑
倒推到函数调用位置(github)
checktime(); // Fail early if deadline has already been exceeded
提示信息为:如果已超过截止日期,则提早失败
继续反推代码
由于(github)交易组装时trx.max_cpu_usage_ms
固定设置的0。所以,判断逻辑出自(github)
// Possibly lower objective_duration_limit to the maximum cpu usage a transaction is allowed to be billed
if( cfg.max_transaction_cpu_usage <= objective_duration_limit.count() ) {
objective_duration_limit = fc::microseconds(cfg.max_transaction_cpu_usage);
billing_timer_exception_code = tx_cpu_usage_exceeded::code_value;
_deadline = start + objective_duration_limit;
}
auto& rl = control.get_mutable_resource_limits_manager();
objective_duration_limit = fc::microseconds( rl.get_block_cpu_limit() );
_deadline = start + objective_duration_limit;
if(now > _deadline) {
// 报异常
}
get_mutable_resource_limits_manager(github)
resource_limits_manager& controller::get_mutable_resource_limits_manager()
{
return my->resource_limits;
}
get_block_cpu_limit(github)
uint64_t resource_limits_manager::get_block_cpu_limit() const {
const auto& state = _db.get<resource_limits_state_object>();
const auto& config = _db.get<resource_limits_config_object>();
return config.cpu_limit_parameters.max - state.pending_cpu_usage;
}
跟进停止
发现是由于合约中使用了std::vector<std::string>
持续存储大量数据,导致单次CPU运算执行耗时持续增长。
结论
当合约中Table单条记录中,如果使用vector
不要存储过多数据,以及数据结构过于复杂的struct
。
如有对应需求,可以拆分子表,然后使用子表记录id索引对应。