https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234754
Bug ID: 234754
Summary: FreeBSD 12.0 ena Network Driver on AWS EC2 Packet Loss
Product: Base System
Version: 12.0-RELEASE
Hardware: Any
OS: Any
Status: New
Severity: Affects Only Me
Priority: ---
Component: kern
Assignee: b...@freebsd.org
Reporter: mwal...@carbonhouse.com
Overview:
Given a single connection transferring anywhere a 100MB file, using scp or just
plain netcat, there will eventually be sufficient packet loss on the server
side such that the connection will stall and eventually reset.
Steps to Reproduce:
1. Boot the standard FreeBSD 12.0 AMI ami-04331586c79df8e01 in us-west-2
2. Attempt to use scp to copy a 100MB file onto the server
Additional Information:
I've been able to replicate this against a bone-stock boot-up of the FreeBSD
12.0 AMI on the following instance+region combinations (every one I've tried
I've been able to replicate it on):
* eu-west-1 / t3.medium
* us-west-2 / t3.small
* us-west-2 / t3.medium
* us-west-2 / t3.large
* us-west-2 / m5.large
* us-west-2 / m5a.large
* us-west-2 / m5a.xlarge
I was able to replicate this using a residental (~50Mbps) connection as well as
another hosted machine at another provider with a 1Gbps connection, though I
cannot replicate this using a machine in the same VPC as the server as the
client.
This really seems like the network driver is just black-holeing packets. I see
no error messages in dmesg, /var/log/, the console, or anything untoward in the
dev.ena.* sysctl values :(
Here's `sysctl dev.ena` from a t3.small instance after a number of tcp
connections experiencing packet loss:
dev.ena.0.admin_stats.no_completion: 0
dev.ena.0.admin_stats.out_of_space: 0
dev.ena.0.admin_stats.completed_cmd: 37
dev.ena.0.admin_stats.sumbitted_cmd: 37
dev.ena.0.admin_stats.aborted_cmd: 0
dev.ena.0.hw_stats.rx_drops: 0
dev.ena.0.hw_stats.tx_bytes: 5681243
dev.ena.0.hw_stats.rx_bytes: 47398849
dev.ena.0.hw_stats.tx_packets: 27863
dev.ena.0.hw_stats.rx_packets: 42686
dev.ena.0.queue1.rx_ring.empty_rx_ring: 0
dev.ena.0.queue1.rx_ring.bad_req_id: 0
dev.ena.0.queue1.rx_ring.bad_desc_num: 0
dev.ena.0.queue1.rx_ring.dma_mapping_err: 0
dev.ena.0.queue1.rx_ring.mjum_alloc_fail: 0
dev.ena.0.queue1.rx_ring.mbuf_alloc_fail: 0
dev.ena.0.queue1.rx_ring.bad_csum: 0
dev.ena.0.queue1.rx_ring.refil_partial: 0
dev.ena.0.queue1.rx_ring.bytes: 33640679
dev.ena.0.queue1.rx_ring.count: 28882
dev.ena.0.queue1.tx_ring.mbuf_collapse_err: 0
dev.ena.0.queue1.tx_ring.mbuf_collapses: 0
dev.ena.0.queue1.tx_ring.bad_req_id: 0
dev.ena.0.queue1.tx_ring.missing_tx_comp: 0
dev.ena.0.queue1.tx_ring.doorbells: 18803
dev.ena.0.queue1.tx_ring.dma_mapping_err: 0
dev.ena.0.queue1.tx_ring.prepare_ctx_err: 0
dev.ena.0.queue1.tx_ring.bytes: 4843829
dev.ena.0.queue1.tx_ring.count: 18812
dev.ena.0.queue0.rx_ring.empty_rx_ring: 0
dev.ena.0.queue0.rx_ring.bad_req_id: 0
dev.ena.0.queue0.rx_ring.bad_desc_num: 0
dev.ena.0.queue0.rx_ring.dma_mapping_err: 0
dev.ena.0.queue0.rx_ring.mjum_alloc_fail: 0
dev.ena.0.queue0.rx_ring.mbuf_alloc_fail: 0
dev.ena.0.queue0.rx_ring.bad_csum: 0
dev.ena.0.queue0.rx_ring.refil_partial: 0
dev.ena.0.queue0.rx_ring.bytes: 13758170
dev.ena.0.queue0.rx_ring.count: 13804
dev.ena.0.queue0.tx_ring.mbuf_collapse_err: 0
dev.ena.0.queue0.tx_ring.mbuf_collapses: 0
dev.ena.0.queue0.tx_ring.bad_req_id: 0
dev.ena.0.queue0.tx_ring.missing_tx_comp: 0
dev.ena.0.queue0.tx_ring.doorbells: 9057
dev.ena.0.queue0.tx_ring.dma_mapping_err: 0
dev.ena.0.queue0.tx_ring.prepare_ctx_err: 0
dev.ena.0.queue0.tx_ring.bytes: 838748
dev.ena.0.queue0.tx_ring.count: 9057
dev.ena.0.admin_q_pause: 0
dev.ena.0.interface_down: 1
dev.ena.0.interface_up: 2
dev.ena.0.wd_expired: 0
dev.ena.0.missing_tx_threshold: 128
dev.ena.0.missing_tx_max_queues: 4
dev.ena.0.missing_tx_timeout: 21474836480
dev.ena.0.keep_alive_timeout: 25769803776
dev.ena.0.wd_active: 1
dev.ena.0.%parent: pci0
dev.ena.0.%pnpinfo: vendor=0x1d0f device=0xec20 subvendor=0x0000
subdevice=0x0000 class=0x020000
dev.ena.0.%location: slot=5 function=0 dbsf=pci0:0:5:0 handle=\_SB_.PCI0.S05_
dev.ena.0.%driver: ena
dev.ena.0.%desc: ENA adapter
dev.ena.%parent:
--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
freebsd-bugs@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-bugs
To unsubscribe, send any mail to "freebsd-bugs-unsubscr...@freebsd.org"
