bgpd: avoid clearing routes for peers that were never established (backport #16271) #16302
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![mergify[bot]](https://avatars.githubusercontent.com/in/10562?s=60&v=4){kind=small}
Under heavy system load with many peers in passive mode and a large number of routes, bgpd can enter an infinite loop. This occurs while processing timeout BGP_OPEN messages, which prevents it from accepting new connections. The following log entries illustrate the issue: >bgpd[6151]: [VX6SM-8YE5W][EC 33554460] 3.3.2.224: nexthop_set failed, resetting connection - intf 0x0 >bgpd[6151]: [P790V-THJKS][EC 100663299] bgp_open_receive: bgp_getsockname() failed for peer: 3.3.2.224 >bgpd[6151]: [HTQD2-0R1WR][EC 33554451] bgp_process_packet: BGP OPEN receipt failed for peer: 3.3.2.224 ... repeating The issue occurs when bgpd handles a massive number of routes in the RIB while receiving numerous BGP_OPEN packets. If bgpd is overloaded, it fails to process these packets promptly, leading the remote peer to close the connection and resend BGP_OPEN packets. When bgpd eventually starts processing these timeout BGP_OPEN packets, it finds the TCP connection closed by the remote peer, resulting in "bgp_stop()" being called. For each timeout peer, bgpd must iterate through the routing table, which is time-consuming and causes new incoming BGP_OPEN packets to timeout, perpetuating the infinite loop. To address this issue, the code is modified to check if the peer has been established at least once before calling "bgp_clear_route_all()". This ensures that routes are only cleared for peers that had a successful session, preventing unnecessary iterations over the routing table for peers that never established a connection. With this change, BGP_OPEN timeout messages may still occur, but in the worst case, bgpd will stabilize. Before this patch, bgpd could enter a loop where it was unable to accpet any new connections. Signed-off-by: Loïc Sang <loic.sang@6wind.com> (cherry picked from commit e0ae285)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Under heavy system load with many peers in passive mode and a large number of routes, bgpd can enter an infinite loop. This occurs while processing timeout BGP_OPEN messages, which prevents it from accepting new connections. The following log entries illustrate the issue:
The issue occurs when bgpd handles a massive number of routes in the RIB while receiving numerous BGP_OPEN packets. If bgpd is overloaded, it fails to process these packets promptly, leading the remote peer to close the connection and resend BGP_OPEN packets.
When bgpd eventually starts processing these timeout BGP_OPEN packets, it finds the TCP connection closed by the remote peer, resulting in "bgp_stop()" being called. For each timeout peer, bgpd must iterate through the routing table, which is time-consuming and causes new incoming BGP_OPEN packets to timeout, perpetuating the infinite loop.
To address this issue, the code is modified to check if the peer has been established at least once before calling "bgp_clear_route_all()". This ensures that routes are only cleared for peers that had a successful session, preventing unnecessary iterations over the routing table for peers that never established a connection.
With this change, BGP_OPEN timeout messages may still occur, but in the worst case, bgpd will stabilize. Before this patch, bgpd could enter a loop where it was unable to accpet any new connections.
This is an automatic backport of pull request #16271 done by Mergify.