tickets #111677
closed
Decommissioning of unused machines
Added by lrupp almost 2 years ago.
Updated 9 months ago.
Category:
Servers hosted in NBG
Description
Just out of interest, what should happen with the existing, but not used machines:
[x] fedora-accounts.infra.opensuse.org (last login from hellcp in at Dec 2 - kept up-to date from other people since than)
[x] fedora-freeipa.infra.opensuse.org (last login from hellcp in May 6 - kept up-to date from other people since than)
[x] identification.infra.opensuse.org (last login from hellcp in Dec 2 - kept up-to date from other people since than)
[ ] new-forum.infra.opensuse.org (last login from others than the people keeping the machine up-to date not detectable) - will keep it, see comment 6
If there is no general objection, the following will happen:
[x] machines will be powered down at the end of June 2022
[x] removed from monitoring
[x] removed from backup
[ ] delete the pillar/id/ files of these machines, but keep the actual salt code (roles etc.)
[ ] machine images (incl. data) will be removed at the end of Dec 2022
- Private changed from Yes to No
lrupp wrote:
- new-forum.infra.opensuse.org (last login from others than the people keeping the machine up-to date not detectable)
I can tell I was last logged in a little more than a year ago. I'm guessing this machine was for the planned move to the new vBulletin. Which is on hold until we get the authentication sorted out. I have no idea when that might be.
Sasi, any comment from your side?
IMHO most of the machines were requested by you (if I remember correctly)?
- Status changed from New to Feedback
They can go, none of them have anything of value on them after the decision to use the current horrible login system was reached, so I don't see a reason to keep them around
IMHO new-forum should stay - in the hope that someone finds time to fix the login ;-)
- Description updated (diff)
Machines decommissioned:
- removed from monitoring
- checked backup: nothing in there
- checked Galera-Cluster & PostgreSQL: no databases found
- ran last update
- machines shutdown
How do we/should we handle:
- Salt certificate: should we delete the key (to avoid that the salt-master is waiting for the offline machines on each global run)?
- Git repo: should we keep the configurations of the machines in the repository (including needed adjustments in the future, if there are changes affecting our CI)?
lrupp wrote:
How do we/should we handle:
- Salt certificate: should we delete the key (to avoid that the salt-master is waiting for the offline machines on each global run)?
Yes, please salt-key -d them.
- Git repo: should we keep the configurations of the machines in the repository (including needed adjustments in the future, if there are changes affecting our CI)?
Usually I'd say delete the pillar/id/ files of these machines, but keep the actual salt code (roles etc.). However, our CI scripts check for unused roles, which makes this question more interesting[tm].
@hellcp: what's your opinion on this?
- Description updated (diff)
- Description updated (diff)
- Description updated (diff)
- Tracker changed from communication to tickets
IMHO new-forum should stay
Is this statement still valid, now that discourse01.i.o.o runs forums.o.o?
On new-forum.i.o.o I only find some PHP/VB5 "stuff".
- Status changed from Feedback to In Progress
- Assignee changed from hellcp to crameleon
- % Done changed from 0 to 50
- fedora-accounts.infra.opensuse.org
- fedora-freeipa.infra.opensuse.org
These machines were not powered on and no Salt keys could be found. I disabled the VM definitions and deleted the disks.
- identification.infra.opensuse.org
This machine was still running and connected with Salt. I shut it down and will proceed with the deletions if there are no complaints within a week.
- Category set to Servers hosted in NBG
@cboltz, @hellcp - can fedora-sso.i.o.o go away as well? The description states it is (was?) a testing machine.
- Status changed from In Progress to Feedback
Removed fedora-sso.i.o.o as there has not been any feedback and the machine seems to be stuck in a GRUB rescue shell since a long time without anyone complaining.
Leaves the question about new-forum.i.o.o. @cboltz
Seems the forums application on new-forum.i.o.o is broken anyways - nginx points to /run/php-fpm/forums.sock, but there is only /run/php-fpm/forums. The last nginx access log entries are from a week ago, some web scraping bot. I shut it down now and if nobody complains within a few days I'll remove it.
crameleon wrote:
IMHO new-forum should stay
Is this statement still valid, now that discourse01.i.o.o runs forums.o.o?
On new-forum.i.o.o I only find some PHP/VB5 "stuff".
Yeah, I don't think we're going back to VBulletin anymore, right?!
- Subject changed from Decomission of unused machines to Decommissioning of unused machines
crameleon wrote:
Seems the forums application on new-forum.i.o.o is broken anyways - nginx points to /run/php-fpm/forums.sock, but there is only /run/php-fpm/forums. The last nginx access log entries are from a week ago, some web scraping bot. I shut it down now and if nobody complains within a few days I'll remove it.
Confirm new-forum.i.o.o can be removed - it was used for testing a fix of the vBulletin database, but we have since migrated off vBulletin.
- Status changed from Feedback to In Progress
- % Done changed from 50 to 60
- % Done changed from 60 to 80
- Status changed from In Progress to Resolved
- % Done changed from 80 to 100
All listed machines have now been taken care of.
Also available in: Atom
PDF