Physical movement of machines preparation

Hello all

With the last night update to discourse I also tested that lists.staging can be reached.
So all tests are working.
For your reference I am attaching the notes from the meeting about the datacenter migration with Lance from OSUOSL.

Action item: Virtualize osgeo6

We have ready lists.staging.osgeo.org, Finished making tests from containers that heavily use the mail server. Being the last test discourse while making the update last night.
The list of all tests can be found here:
https://gitea.osgeo.org/sac/osgeo9/wiki/mailserver-container#test-mails

There are other services in osgeo7 and osgeo8, specially tracsvn that use the mail server heavily. that one was not tested.

The final steps of this action would be:

  • Shutdown osgeo5/backup (the mail server backup)

  • This can be done anytime as we are already running a backup on osgeo7

  • On osgeo9: Make a final rsync with osgeo6

  • disable the cronjob

  • On osgeo9: Add to sites-enabled lists.osgeo,org.conf

  • On osgeo9: Change mm_cfg.py DEFAULT_EMAIL_HOST and ``DEFAULT_URL_HOST

  • Change the DNS settings of lists.osgeo.org on Cloudflare

  • If everything goes well

  • Final cleanup of cloudflare

  • on osgeo7: (osgeo6-buster) modify the cronjob to update from osgeo9 instead of osgeo6

  • Otherwise:

  • Restore original configuration on cloudflare

  • Restore the cronjob

  • Restore the mm_cfg.py file

When to do this final step will depend on how do we tackle the physical movement:



Hostname



Model



Migrate?



osgeo-backup (osgeo5)



Has only 4 cores and 16 GB ram



no



osgeo3



Dell R710



yes?



osgeo4



Dell R710



no?



osgeo6



Supermicro X10DRL-i



yes



osgeo7



Supermicro X11DDW-L



yes



osgeo8



HP DL360 G6



yes



osgeo9



HP DL360 G6



yes

We are only migrating: osgeo6, osgeo7, osgeo8 and osgeo9
but osgeo6 is going to be reformatted so we can use it for virtualization.
During the physical movement services on the moved machine will be shutdown.

My proposed way of doing the migration:

  • Migrate osgeo9
  • this would keep alive tracsv
  • when turning back on the services will continue running normally
  • Do the final steps, mentioned above, for using the containerized mail server on osgeo9
  • Migrate osgeo7
  • Migrate osgeo8
  • Migration of osgeo6:
  • A new disk needs to be bought and installed
  • Lance will do us the favor of installing and formatting and initial installation of operative system for virtualization
  • He can do on the current datacenter or the new datacenter
  • Migrate contents of osgeo3 and osgeo4
  • staging can go directly to osgeo6
  • The rest we can decide where to put the conainers
  • Shutdown osgeo3 and osgeo 4

Please let me know your comments

Regards
Vicky

(attachments)

Notes - FW_ OSGEO Datacenter Migration Planning.pdf (64.8 KB)

On Sun, Sep 14, 2025 at 02:19:45PM -0600, Vicky Vergara wrote:

We have ready lists.staging.osgeo.org, Finished making tests from
containers that heavily use the mail server. Being the last test discourse
while making the update last night.
The list of all tests can be found here:
Making sure you're not a bot!

That page is not very clear to me, doesn't say what was tested exactly.

The services provided by current OSGeo6 are:

  - Mail Submission Agent
    SAC:Message Submission Agent - OSGeo
    To test:
      - ability to SEND email from an @osgeo.org
        (@staging.osgeo.org while staging)
        using the OSGeo UserID credentials.
      - Mail authentication (SPF/DMARC/DKIM)

  - Mail Transport Agent
    SAC:Message Transport Agent - OSGeo
    To test:
      - Ability to receive mail sent to an @osgeo.org alias
        (@staging.osgeo.org alias while in staging)
      - Mail evaluation (DMARC/SPF/DKIM)
      - Spam handling (graylists etc.)

  - Mailing lists service
    SAC:Mailing Lists - OSGeo
    To test:
      - Ability to create a list, subscribe, send to and
        receive mail from the list, via mail.
      - Ability to see archives and login via web form.

  - The www.tilecache.org website
    Making sure you're not a bot!
    To test:
      - Ability to still reach it (unless it's dismissed)

  - MySQL ?
    This is found on the wiki but it isn't clear if the service if needed
    can we drop it safely ?

There are other services in osgeo7 and osgeo8, specially tracsvn that use
the mail server heavily. that one was not tested.

The Trac mail setup is a bit confused at the moment in that some
projects use "localhost" as the mail server and other
"mail.osgeo.org", I've just filed a ticket to simplify it:

The tracsvn postfix service doesn't have any special configuration to
be trusted more than a normal SMTP connection so the ability to
connect to port 25 of the new IP and send mail via it should be enough
as a test.

The final steps of this action would be:
- Shutdown osgeo5/backup (the mail server backup)
  - This can be done anytime as we are already running a backup on osgeo7

Better do this now, if not needed, less confusion later.

- On osgeo9: Make a final rsync with osgeo6

My understanding is that the current "rsync" doesn't sync all the data
on the disk, but just a subset, could you tell which subset of data
and configurations are synchronized ?

In particular, do we get queued email ?
At the moment osgeo6 has 2079 queued emails.

I think we need to figure out how to ensure no queued mail gets lost,
so a way to have the NEW service process the OLD queues, or keep the
old service alive until it's finished with those queues while not
accepting new ones...

Is there anything in the new container that should be DIFFERENT
from the old container ? Otherwise why not just rsync everything ?

--strk;

  Libre GIS consultant/developer :trumpet:
  strk's services