Peers section logging


I’ve got a weird problem with haproxy peering. Everything is working fine, haproxy instances are syncing stick tables between them correctly up until some point where it stops completely and the proto_err counter starts to go up. There’s not a single error pushed to the logs. Is there a way to get any logs about what’s going on with the haproxy peering?

Here’s partial output from the peers state:

0x55939dff3380: [12/Dec/2020:09:24:58] id=mypeers state=0 flags=0x3 resync_timeout=22d7h task_calls=1093943
  0x55939dff92f0: id=haproxy-1(remote) addr=X:1024 status=ESTA reconnect=4s confirm=0 tx_hbt=773183 rx_hbt=771493 no_hbt=18 new_conn=10136 proto_err=4395
        flags=0x20000000 appctx:0x7ffb4801fea0 st0=7 st1=0 task_calls=2186 state=EST
        last_local_table:0x55939e007260 id=jitsi-meet local_id=1 remote_id=1
        shared tables:
          0x55939e007260 local_id=1 remote_id=1 flags=0x0 remote_data=0x80001
              last_acked=0 last_pushed=46352 last_get=0 teaching_origin=46269 update=46352
              table:0x55939dffcc20 id=jitsi-meet update=46352 localupdate=46352 commitupdate=46352 syncing=0

Please file a bug at:

@lukastribus I will, but we have just upgraded to 2.3.0 and we need to have a way for monitoring the protocol errors or the peering state in general. It seems that it’s possible to get trace logs out of the peers module:

Does anyone know how to enable tracing for the peers module? When I execute “show trace” it doesn’t list peers as one fo the trace sources:

root@haproxy-1:/# echo "show trace" | socat stdio UNIX-CONNECT:/var/run/hapee-lb.sock
Supported trace sources and states (.=stopped, w=waiting, R=running) :
 [.] h2         -> none [drp 0]  [HTTP/2 multiplexer]
 [.] fcgi       -> none [drp 0]  [FastCGI multiplexer]
 [.] stream     -> none [drp 0]  [Applicative stream]
 [.] h1         -> none [drp 0]  [HTTP/1 multiplexer]