Dealt with closing connections to redis via haproxy.
Added to gitlab-admin the ability to specify environment lists for variables. During application, a separate variable is created for each element of the list.
Optimized the acme.sh installation state, now each account has its own crown and has a separate root folder.
Received the long-awaited hoodies and t-shirts with our wonderful cat logo.
Fixed a bug in the salt template of the project that broke the sending of alerts from pipelines via salt-ssh.
Disks whose path contains “kubernetes.io” with disk_alert were excluded, because in this way pvc volumes on kubernetes cluster nodes were monitored, which is unnecessary.
Added “normality” of status 2 to status 4 to the PXC cluster status check utility, otherwise critical errors appeared on the monitoring dashboard at the moment when a cluster node became a donor for another node.
Added the ability to set severity for jobs in accounting. In this way, you can raise the severity from the default major, for which alerts are not sent to chats, to, for example, critical, so that the alerting works.
Investigated the possibility of attracting customers through Upwork. Conclusion: even though Upwork has an “agency” model and potentially interested clients in our type of services, the procedures are so regulated and complicated that it is not possible for us to use it. But it was an interesting experience researching niches :).