CODEWITHSUNDEEP
tcpicmpmtumss
yaodongen
yaodongen

Reputation: 31

The possible reasons for not receiving ICMP messages

We encountered an issue where we upload customer files to AWS S3. It lasted for about 10 minutes, with all uploads failing, then it recovered on its own. During troubleshooting, we found that it might be due to a small MTU configuration in the network path, because running sudo ping -D -s 1000 xxxxx.com to the server returned an error. But -s 500 returned ok. The ping returned to normal after 10 minutes.

$ ping -D -s 1000  xxxxx.com
PING xxxxx.com (39.156.66.10): 1000 data bytes
ping: sendto: Message too long

$ping -D -s 500  xxxxx.com
PING xxxxx.com (39.156.66.10): 500 data bytes
508 bytes from 39.156.66.10: icmp_seq=0 ttl=64 time=0.377 ms
508 bytes from 39.156.66.10: icmp_seq=1 ttl=64 time=0.664 ms

However, as far as I know, when the MTU suddenly becomes smaller, a type=3, code=4 ICMP message PTB (Packet Too Big) should be received, and TCP/IP should adjust the size and still upload successfully. Some people suggest that the issue is caused by the load balancer dropping the ICMP messages, but this doesn't seem quite right. It has been a long time, and I can't provide any more information. I would like to know the possible reasons. Thank you very much.

Upvotes: 0

Views: 31

Answers (0)

Related Questions