Wed, Oct 21
This has been done several times over, and so this ticket isn't actually that useful.
Thu, Oct 1
Tue, Sep 29
Mon, Sep 28
Fri, Sep 25
I believe this is complete, any further patches should point at the parent task T258729
Sep 24 2020
Sep 23 2020
These all pass Python 3 tox and have no patches for automated porting, so seem ready for the Python 3 future. They are low priority to touch at all so I'll let the lie for now.
Sep 22 2020
Sep 21 2020
This has been approved in the team meeting and followed up on IRC. Let us know if there's anything that needs to be done further.
I see the patch is already merged, @nskaggs please test icinga command and followup so we can close ticket. Thanks!
Sep 16 2020
We shall have to destroy this and recreate it as buster which should be relatively painless. We actually do need this since we need "secrets" for this cluster of boxes afaik.
Sep 14 2020
Sep 11 2020
Sep 10 2020
Sep 9 2020
Sep 2 2020
Done and done.
Sep 1 2020
Some notes about the current progress:
- Imports seem 99% good.
- Massimport miniscript deals with the manual cases
- there is a hack in the interface automation script to deal with some mis-netmasking on some extra ip addresses on aqs*,restbase*,sessionstore*
Aug 31 2020
Aug 28 2020
Aug 21 2020
DNS vs host mismatch:Host ganeti5001 has inet6 2001:df2:e500:101:d294:66ff:fe81:9090/64 while DNS has 1H IN AAAA 2001:df2:e500:101:10:132:0:21 similar for ganeti5002 and ganeti5003
FAILED not in puppetdb:Found origin 21.64.10 Missing line: 125 1H IN PTR an-presto1004.eqiad.wmnet. Found origin 18.104.22.168.22.214.171.124.0.0.0.0.0.2.6.2 Missing line: 126.96.36.199.188.8.131.52.184.108.40.206.0.1.0.0 1H IN PTR an-presto1004.eqiad.wmnet. Missing line: an-presto1004 1H IN A 10.64.21.125 Missing line: an-presto1004 1H IN AAAA 2620:0:861:105:10:64:21:125 Found origin 16.192.10 Missing line: 47 1H IN PTR wtp2005.codfw.wmnet. Missing line: wtp2005 1H IN A 10.192.16.47
Aug 20 2020
Aug 10 2020
I have fixed this problem and placed a long term fix in place.
Aug 4 2020
update on this project:
Jul 31 2020
I have dumped and reloaded from a more recent dump (2020-07-31) and rerun import with preserving all IP addresses in the 'bindings' list instead of just primary IPv4 and IPv6.
Jul 30 2020
In terms of the extra IP addresses, the import only looks at the ip/ip6 fields and not the binding fields. It would be "relatively" easy to make it look at interface bindings instead but it would mean restructuring some things. A big problem is that the names aren't stored in puppetdb at all, and would have to be manually added to those IP addresses anyway, which makes me think that extra things like these should be handled in a manual fashion or at least separately.
This sounds completely doable, and not to big of a deal.
Jul 29 2020
Oh also note recent update fixes devices which use previously reserved ip addresses now set the state to active and clear the comment if the comment is 'reserved for infra'
I have dumped and reloaded, the results are similar but now all VMs have their interfaces set to the virtual type since the API explodes if you don't.
Jul 28 2020
Jul 27 2020
Yep, will do. Moving that forward to asap.
Jul 23 2020
Jul 22 2020
Jul 13 2020
Jul 6 2020
This is complete.
Jun 30 2020
Jun 23 2020
Thank you, I'll take care of it.
Jun 17 2020
It should be straight forward to add a filter as you suggest although it's notable that the puppetdb proxy already gives us the short hostname so it'll have to altered very slightly but nbd i think.
Nice, this is what we pretty much had in mind, although in the future of course if we add more prefixes or change them we'll have to reserve them manually or whatever.