[SOLVED] error spam: [udp: tx_sock_error, subsys=22, recv_err=-1, recv_errno=90] ?

S

scyto

Active Member
Aug 8, 2023
398
72
28
I am seeing the syslog on all 3 of my nodes being spammed as follows every 30 seconds.
I am unclear if the error is caused by the two MTU detections or not.
Is this something i can ignore or need to get to the bottom of?
Code: 
....
Aug 16 15:28:09 pve2 corosync[1056]:   [KNET  ] pmtud: PMTUD completed for host: 1 link: 0 current link mtu: 1397
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] pmtud: Starting PMTUD for host: 3 link: 0
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] udp: tx_sock_error, subsys=22, recv_err=-1, recv_errno=90
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] udp: detected kernel MTU: 1500
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] pmtud: PMTUD completed for host: 3 link: 0 current link mtu: 1397
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] pmtud: Starting PMTUD for host: 1 link: 0
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] udp: tx_sock_error, subsys=22, recv_err=-1, recv_errno=90
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] udp: detected kernel MTU: 1500
Aug 16 15:28:39 pve2 corosync[1056]:   [KNET  ] pmtud: PMTUD completed for host: 1 link: 0 current link mtu: 1397
.....

corosync seems to be working...
 
scyto said:
This seems to be result of over aggressive logging, these are completely normal and in this https://github.com/corosync/corosync/issues/716 it references a pull that moves these to trace level messages.

@t.lamprecht sorry to tag you, but this issue makes the logs SUPER noisy, is this on the teams radar (this was happening before i had custom kernel).
Click to expand...
We're not actively planning on backporting those patches for knet and then the integration in corosync, but once the next knet/corosync versions are bumped we normally update our packages to that relatively soon.
 
  • Like
Reactions: scyto
You must log in or register to reply here.
Top Bottom