Коллеги, просто для истории - фиксирую текущее состояние...

Colleagues, just for the sake of history - I fix the current state of the infrastructure of key services and processes at our university. To compare with what will remain here in six months or a year. By all indications, they decided to completely destroy it and replace it with crutches 10 years ago, without reservation, without discovering services, without CI / CD, apparently only to take it away from those who put their hearts into the technological development of our university (namely our , we all studied and grew up in it). This especially looks paradoxical if there is a transfer plan approved by the Vice-Rector in support of all existing systems and technologies with documentation and training, with a guarantee of operability, since we check with hundreds of tests.

It is unfortunate that one of the best infrastructures (unfortunately, it worked on not very good hardware and with not very attentive administrators), which we represented even at business forums, will turn into rags. Because "technology is not needed for a pedagogical university" (quote from one of the leaders). Mark my word - in a year there will be a swamp, but, fortunately, we will no longer be connected with this. We have done and are doing all the necessary actions for the transfer of systems, but all our best practices and experience can go to the trash, and you can find out the reasons yourself, this is not a secret.

We are changing and starting to work with students and teachers, those people who have sufficient competencies and a vision for the future to accept and create new things. And the informatization department in its current state, unfortunately, has become an antipatter in my eyes - sheer lies, intrigues, secret operations, a lot of lost papers and promises and a complete unwillingness to respect employees and listen to questions that are not always pleasant. This is the road to nowhere.

So it was:
All subsystems were transferred to official or specially prepared docker containers, recovery time on failure decreased from tens of minutes to several seconds, a hosting model for containers was implemented (by the way, there was not a single problem with docker hosting, unlike the old chroot, but not only with hosting, docker is VERY stable).
There is an infrastructure for distributed monitoring (Telegraf + InfluxDB / Grafana) and service discovery (Consul), comprehensive monitoring of the availability and functioning of subsystems, scenarios of reaction to anomalies (kapacitor)
Collaboration platform for common workspaces, process automation (Alfresco ECM) launched and stable
A full-fledged CI / CD was built on Gitlab and tools for updating the debugging and production versions of systems (both for web systems and container packaging), assembly and testing at all stages are automated
Two buses are implemented: for distributed applications (RabbitMQ) and for data streams (Apache QPID), the ability to aggregate data in a heterogeneous environment is available
The account directory is implemented on the ReOpenLDAP cluster (we found and helped fix several errors in the system code) from 3 nodes, an experimental domain was launched on SMB4 and Apereo CAS (OAuth)
The mail system is made with full redundancy SMTP (exim4), IMAP (dovecot), MX (exim4 + sa), a web interface with a beautiful design and functionality.
Backup to Bacula, manage data sets through puppet.