"""
基本思想：
    用队列优化Bellman-Fold算法，长得有点像堆优化的Dijkstra算法
    在a->b的边中，Bellman-Fold算法的更新公式为：dist[b] = min(dist[b], dist[a]+weight),若想dist[b]变小，则dist[a]一定需要变小
SPFA算法模板：
def spfa():
    dist = [float("inf") for i in range(n+1)]
    dist[1] = 0
    queue = [1]
    st[1] = True    # ！！！注意，st数组用来标记节点i是否在queue中

    while queue:
        t = queue.pop(0)
        st[t] = False
        if t not in graph: continue

        for weight, node in graph[t]:
            if dist[node]>dist[t]+weight:    # !!!注意：判断dist[t]是否变小，变小则更新距离，同时将node添加到队列中
                dist[node] = dist[t] + weight
                if st[node]==False:    # node不在队列中
                    queue.append(node)
                    st[node] = True

    if dist[n]>=float("inf")/2:
        return -1
    else: 
        return dist[n]

"""

def spfa():
    dist = [float("inf") for i in range(n+1)]
    dist[1] = 0
    queue = [1]
    st[1] = True    # ！！！注意，st数组用来标记节点i是否在queue中

    while queue:
        t = queue.pop(0)
        st[t] = False
        if t not in graph: continue

        for weight, node in graph[t]:
            if dist[node]>dist[t]+weight:
                dist[node] = dist[t] + weight
                if st[node]==False:    # node不在队列中
                    queue.append(node)
                    st[node] = True


    if dist[n]>=float("inf")/2:
        return -1
    else: 
        return dist[n]

# 输入示例
if __name__=="__main__":
    n, m = map(int, input().split())
    st = [False for i in range(n+1)]    # ！！！注意，st用来标记节点i是否在queue中

    # 带权图的存储模板
    graph = {}
    while m:
        x, y, z = map(int, input().split())
        if x not in graph:
            graph[x] = [(z, y)]
        else:
            graph[x].append((z, y))
        m -= 1

    res = spfa()
    print(res) if res!= -1 else print("impossible")