In our K8 cluster, we use haproxy app for connecting to Galera cluster.
Our haproxy.cnf file looks like
global
maxconn 2048
external-check
stats socket /var/run/haproxy.sock mode 600 expose-fd listeners level user
user haproxy
group haproxy
defaults
log global
mode tcp
retries 10
timeout client 30000
timeout connect 100500
timeout server 30000
frontend mysql-router-service
bind *:6446
mode tcp
option tcplog
default_backend galera_cluster_backend
# MySQL Cluster BE configuration
backend galera_cluster_backend
mode tcp
option tcpka
option mysql-check user haproxy
balance source
server pitipana-opsdb1 192.168.144.82:3306 check weight 1
server pitipana-opsdb2 192.168.144.83:3306 check weight 1
server pitipana-opsdb3 192.168.144.84:3306 check weight 1
Dockerfile for creating haproxy image
FROM haproxy:2.3
COPY haproxy.cfg /usr/local/etc/haproxy/haproxy.cfg
In my Galera nodes, I get constant warning in /var/log/mysql/error.log
2021-12-20 21:16:47 5942 [Warning] Aborted connection 5942 to db: 'ourdb' user: 'ouruser' host: '192.168.1.2' (Got an error reading communication packets)
2021-12-20 21:16:47 5943 [Warning] Aborted connection 5943 to db: 'ourdb' user: 'ouruser' host: '192.168.1.2' (Got an error reading communication packets)
2021-12-20 21:16:47 5944 [Warning] Aborted connection 5944 to db: 'ourdb' user: 'ouruser' host: '192.168.1.2' (Got an error reading communication packets)
I had increased max_packet_size to 64MB and max_connections to 1000.
When I take a tcpdump from galera node :
Frame 16: 106 bytes on wire (848 bits), 106 bytes captured (848 bits)
Linux cooked capture
Internet Protocol Version 4, Src: 192.168.1.2, Dst: 192.168.10.3
Transmission Control Protocol, Src Port: 62495, Dst Port: 3306, Seq: 1, Ack: 1, Len: 50
Source Port: 62495
Destination Port: 3306
[Stream index: 2]
[TCP Segment Len: 50]
Sequence number: 1 (relative sequence number)
[Next sequence number: 51 (relative sequence number)]
Acknowledgment number: 1 (relative ack number)
0101 .... = Header Length: 20 bytes (5)
Flags: 0x018 (PSH, ACK)
000. .... .... = Reserved: Not set
...0 .... .... = Nonce: Not set
.... 0... .... = Congestion Window Reduced (CWR): Not set
.... .0.. .... = ECN-Echo: Not set
.... ..0. .... = Urgent: Not set
.... ...1 .... = Acknowledgment: Set
.... .... 1... = Push: Set
.... .... .0.. = Reset: Not set
.... .... ..0. = Syn: Not set
.... .... ...0 = Fin: Not set
[TCP Flags: ·······AP···]
Window size value: 507
[Calculated window size: 64896]
[Window size scaling factor: 128]
Checksum: 0x3cec [unverified]
[Checksum Status: Unverified]
Urgent pointer: 0
[SEQ/ACK analysis]
[Timestamps]
TCP payload (50 bytes)
[PDU Size: 45]
[PDU Size: 5]
MySQL Protocol
Packet Length: 41
Packet Number: 1
Request Command SLEEP
Command: SLEEP (0)
Payload: 820000008000012100000000000000000000000000000000...
[Expert Info (Warning/Protocol): Unknown/invalid command code]
[Unknown/invalid command code]
[Severity level: Warning]
[Group: Protocol]
MySQL Protocol
Packet Length: 1
Packet Number: 0
Request Command Quit
Command: Quit (1)
Here 192.168.1.2 is a K8 worker node and 192.168.10.3 is the galera node.
When I connect our applications in K8, we can access to applications, but when we try to edit, we get stuck.
Any suggestion to fix this?
