很久没有发贴。贡献一小细节以获取下载资格：）
Cisco 路由器的出接口是GE, show int 看到很多packets drop信息。 show buffers 有大量的failures. 同时cpu接近95%。
考虑到isp只有可怜的20M,大概不大可能因为只是转发就导致cpu如此之高。在查看了有大量free memory之后，遂决定增加buffers的数量。如下所列：cisco router处理buffers的顺序是小的没了就用大的，一直到huge都没了就丢弃。所以要斟酌考虑增加，我从small到huge都增加了数百到几十个不等。命令很简单：
buffers small max-free 175
buffers small permanent 125
buffers small min 50 同样middle, big 以此类推。 通常min 是permanent 的25%，max大概是130%左右。但是也不是绝对的。经过大幅增加buffers, failures 显著降低，cpu 占用率保持在50%左右。
Router233#show buffers
Buffer elements:
1239 in free list
197699403 hits, 0 misses, 1242 created
Public buffer pools:
Small buffers, 104 bytes (total 343, permanent 300, peak 757 @ 7w0d):
339 in free list (80 min, 500 max allowed)
142250002 hits, 13052 misses, 18375 trims, 18418 created
1954 failures (0 no memory)
Middle buffers, 600 bytes (total 200, permanent 200, peak 208 @ 17:42:21):
199 in free list (50 min, 260 max allowed)
52018712 hits, 8413 misses, 14036 trims, 14036 created
2231 failures (0 no memory)
Big buffers, 1536 bytes (total 150, permanent 150, peak 182 @ 17:42:30):
150 in free list (50 min, 200 max allowed)
10504060 hits, 19328 misses, 5487 trims, 5487 created
15263 failures (0 no memory)
VeryBig buffers, 4520 bytes (total 60, permanent 60):
60 in free list (10 min, 100 max allowed)
7049 hits, 8479 misses, 479 trims, 479 created
8479 failures (0 no memory)
Large buffers, 5024 bytes (total 20, permanent 20, peak 21 @ 17:39:39):
20 in free list (5 min, 30 max allowed)
744 hits, 7735 misses, 2560 trims, 2560 created
7735 failures (0 no memory)
Huge buffers, 18024 bytes (total 10, permanent 10, peak 15 @ 17:39:54):
10 in free list (5 min, 30 max allowed)
2238 hits, 6383 misses, 5612 trims, 5612 created
5503 failures (0 no memory)
发表于 2019-1-11 23:29:47
59
来自手机