Console rendering invalid characters on RHEL 8

[ncstate-mafields]

We recently started upgrading our servers to RHEL 8, and while everything seems to be functional, we are getting invalid characters rendered to the terminal when we use the CLI Console:

[root@dbmgr-dev ~]# replication-manager-cli console
%p1%dGQ������������������������������������������������������

 Replication Manager Client | Group: dev-cluster |  Mode: Manual �����������������������                                                                                                                                                                                                                                                                                                                                                          
Host   Port          Status   Failures   Using GTID         Current GTID           Slave GTID             Replication Health  Delay  RO���������������������                                            �����������������������>   
10.36.21.123   3306          Master          0           No 2-1451829552-10767,7-1451862526-13 2-1451843462-9911,7-1451862526-13                      Master OK      0 OFF�������������     
10.32.21.123   3306           Slave          0    Slave_Pos   2-1451843462-11622   2-1451829552-10767                     Running OK      0  ON

Updates to the terminal are not rendered correctly, and exiting the Console requires a reset command to restore the terminal to normal.

I've tried both the v2.2.20 and v2.3.20 versions and both exhibit the same issue.

I've also verified the issue occurs in CentOS 8 Stream, to rule out any RHEL specific shenanigans.

[svaroqui]

Thanks for reporting we will try to improve on this when one get time in the team

[svaroqui]

Can you please just let us know the name of your cluster so that we can reproduce that bad printing ?

[ncstate-mafields]

Here's the cluster.d/dev-cluster.toml:

[dev-cluster]
title = "Dev Cluster"
db-servers-hosts = "10.36.21.123,10.32.21.123"
db-servers-prefered-master = "10.36.21.123"
db-servers-ignored-hosts = ""
db-servers-connect-timeout = 5
db-servers-read-timeout = 15
db-servers-credential = "[REDACTED]"
db-servers-tls-ca-cert = ""
db-servers-tls-client-cert = ""
db-servers-tls-client-key = ""

replication-credential = "[REDACTED]"
replication-use-ssl = false
replication-master-slave = true

proxy-servers-read-on-master = "true"

[ncstate-mafields]

It looks like this is an issue in the underlying termbox-go dependency due to an update to ncurses terminfo format: https://github.com/nsf/termbox-go/issues/185

I think the dependency is version-locked to a few commits before the fix was pushed. Updating the termbox-go dependency should fix the issue.

Current version locked commit: https://github.com/nsf/termbox-go/commit/88b7b944be8b Commit with fix for new ncurses format: https://github.com/nsf/termbox-go/commit/08476232a995

https://github.com/signal18/replication-manager/blob/ec79c78222eac5da21add8f153d3ada40b01e77b/go.mod#L65

In the meantime, prefixing the replication-manager-cli console command with TERM=xterm works around the issue. The default TERM environment value in RHEL 8 of xterm-256color should reproduce the issue.

[svaroqui]

Hi thanks a lot for reporting and digging in the code for a workaround and a fixe would like to make a pull request for it or should we proceed with the change ?

[ncstate-mafields]

I'm not very fluent in the language, so I'll leave the actual change to the experts :)

[svaroqui]

Please reopen if not fixed