Attendees: jtl@, rscheff@, tuexen@, rrs@
Discussion
PR 252913: tuexen@ got a tracefile and some information from Netflix. This impacts stable/12. rrs@ will commit an update to RACK in main and MFC to stable/13. We don't plan to fix this bug in stable/12.
- Actually, we probably should deprecate (and even remove) RACK code from stable/12. However, we do think we can maintain RACK code throughout stable/13 support cycle.
- Will update tcp(4) and tcp_bbr(4) man pages, and write a new man page for RACK, to note whether stacks are experimental or supported.
D23230: ECN++, rscheff@ updated the sysctl interface so sysctls are not checked in fast path. Needs review.
D28702: Closed.
D29835: Closed.
Lost retransmission detection: D28931 is an idea for doing SACK loss recovery in the base stack. (Documented in an I-D.) We discussed the idea further. rscheff@ will modify code to include a sysctl to enable/disable the new behavior.
- Will be off by default.
- Consensus: add socket option to control on a per-socket basis and add some basic statistics. rrs@ will review. The plan is to commit this before the next meeting.
D29515: rscheff@ committed change to use jenkins hash with random salt and allow an administrator to re-salt with a sysctl. rscheff@ did MFC to stable/13. Closed.
- tuexen@ committed support for TCP over UDP to main. He plans to MFC to stable/13. tuexen@ did this. Closed.
- RACK, BBR, etc. will be updated. rrs@ will commit this week.
- Plan is to MFC to stable/13. See above.
D29441, D28822: PRR-related reviews still open. No plans to MFC D28822.
254725: LRO panic. Peer ACKs up to last data byte. Peer SACKs FIN - see reconstructed trace from in-core packet headers PR254725.trc. snd_una is correct, but send buffer is empty. Still no progress on reproduction. Randall suggested changing the code to wait to send FIN until there was nothing outstanding (see D29565). This change will delay the remote side recognizing session shutdown, which might impact applications which take action when a session closes. General consensus is that it would be better to find and fix the bugs.
- User did not feel that they could afford downtime to reproduce. Will need to await a reproduction.
Follow up on the RST issue reported by NFS developer: We send a challenge ACK. Remote side sends a RST. We don't process the RST. rscheff@ proposed D29690. tuexen@ and jtl@ have a small concern that we should do more work on finding the root cause before we commit a fix. rscheff@ will debug further on a machine seeing this problem.
- Will ask Rick to reproduce and help us figure out the kernel state when we are rejecting the RST.
Next meeting: 20 May @ 1400 UTC