Support on installation of new production environment

[EddyCatt]

OAW is going to prepare a new production environment and they are asking for our support on how to set the IT infrastructure (i.e. to change the OS from CentOS 8 to Ubuntu 20.04...)
Contact reference is Christian Balluch [email protected]

[randomorder]

I had a quick chat with the customer. This is what they suggest:

• they set up the machine with a different distribution (Rocky Linux, very similar to CentOS). Fine by me
• we perform the installation on the system (there are installation instructions available)
• they handle SSL termination at the proxy level. No need to configure SSL on the machine. Fine by me

waiting to have a quick sync with you @EddyCatt and we should be ready to go and ask for the machine to be created.
In terms of effort and we are a bit tight on the active contract. There is only one day left

[randomorder]

Note for the devops: As part of this work also create the missing deployment diagram for both DEV and PROD
https://docs.google.com/drawings/d/110vz4JVLV05rcSLW4sr11J6LXwdKhmqU8L5FnTHsYYc/edit

Project Code is OAW: C190-OAW-2021-GS_Sub

[EddyCatt]

Waiting to talk with the Client about how to manage the support, probably on another contract

[EddyCatt]

OAW is trying to find extra-budget, outside of the former and current contracts, to cover the two days required for the support

[EddyCatt]

A new contract has been opened for this issue: C190-OAW-2021-GeoNode+Airflow
It could be carried out after final fixing of current bug (geosolutions-it/oaw-georectify-process-qgis#19)

[randomorder]

Clone od the DEV Documentation here to be updated

We're targeting end of this week for the deploymeny @lpasquali

[lpasquali]

@randomorder I have a problem and I need to reach the customer:
I can't do any docker pull as docker looks like unreachable from the new server "woldan3"

Using default tag: latest
Error response from daemon: Get "https://registry-1.docker.io/v2/": net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
[root@woldan3 geosol]#

[lpasquali]

@randomorder we should be unblocked

[lpasquali]

@randomorder I finalized an http only installation and airflow, I will finish doc and diagram now
I applied our checklist and everything checks out

[lpasquali]

I updated documentation:
https://docs.google.com/document/d/1MAvNa3dQ-fXs03fb441Cp49CgiVawMEvgbyJs-fxXTM/edit#heading=h.3dy6vkm
and created one more diagram for OAW Airflow
https://docs.google.com/drawings/d/1v2PZ_BeiEzDKUpIE958WrtsT8cn20-mCttpMJhbOOyY/edit
on the latter I think I put the arrows in a way the make sense on how services interact, but if someone could review that diagram it would be better

[randomorder]

I don't have access, use project directory for the docs. Let us know when is ready for review and testing

[lpasquali]

@randomorder I tried both 2021 project folders for OAW, I could put the new diagram I made there, but there is no trace of the pristine diagram you linked, the dev installation document and the cloned prod installation document I compiled for this work, could you please point me to the correct drive folder, so I will use that?

[randomorder]

same name as the label attached to the ticket as usual
you already ahve access

[lpasquali]

@randomorder done https://docs.google.com/drawings/d/1v2PZ_BeiEzDKUpIE958WrtsT8cn20-mCttpMJhbOOyY/edit

[randomorder]

This is ready for testing @EddyCatt

[EddyCatt]

Great, thanks !

[lpasquali]

@randomorder @EddyCatt a reminder which may save some time:
this new server is closed to the internet (woldan2 is not), it can be browsed only through OAW vpn

[EddyCatt]

After testing, the system seems to work, thx @randomorder & @lpasquali !
However, a db dump would have to be made to import layers, maps and geostories from the old server to woldan3

[randomorder]

a db dump would have to be made to import layers, maps and geostories from the old server to woldan3

thanks @EddyCatt we're aiming next week for this operation

[lpasquali]

@randomorder I was able to import everything after some more additions to the "find" migration agreed during our sync:

• after using a script with a find placing the tiffs in the right place, I missed still 6Gb of raster which was not migrated by the geonode backup, I added it using tar - anymore layer that was not fixed by the find, was fixed by this operation.
• after this geoserver login with oauth didn't work anymore, it was due to geonode's various oauth tokens which needed to be cleabed up otherwise the sequence ids of such tables in pg would result in duplicate tables, fixed.
  I then checked that layer preview did not return errors for all layers, it did not except for one with this wms error:

geonode:AC08769695- Content is not allowed in prolog. ???

but this layer is not working as well in dev

[randomorder]

good thanks, I'll let @EddyCatt decide how handle the broken(?) layer. Besides this the task is completed, right @lpasquali ?

[EddyCatt]

Thanks mille @lpasquali ! I'll test it and I'll get back to you.

[lpasquali]

@randomorder by comparison data looked like identical in dev and prod after the work of friday

[EddyCatt]

After testing, I've found 2 issues:

• Getinfo feature is not working on Map view, while it's doing fine on Layer view
• Geostories have not been imported, there were 3 in the development environment

[EddyCatt]

SFTP user needs to have rights to write for the Airflow root folder

[randomorder]

can you fix the system today @lpasquali ?

[lpasquali]

@EddyCatt permissions are aligned as they were in the old server

[randomorder]

can we close this @EddyCatt ?

[EddyCatt]

Unfortunately not. @mattiagiupponi is trying to connect to OAW's VPN to sort out the GeoStories issue but it seems not to work fine.

[lpasquali]

@randomorder as you know @EddyCatt asked me to check why a temporized layer was not working at all.
Disk quota was not working, it said the db password was incorrect, but I actually checked that while importing stuff the first time.
Long story short there is two issues:

• only 4 (actually 3, geonode:all_railway_views_ts was added as was the tiled temporized layer not working) tiled layers, while on the old server are 164
• styles and other things in woldan3 datadir point to woldan2 urls, this is the main problem for layers not working.

[randomorder]

• only 4 (actually 3, geonode:all_railway_views_ts was added as was the tiled temporized layer not working) tiled layers, while on the old server are 164

• You can add the tile layers manually or use a script Script for updating layers. scripts#25

• styles and other things in woldan3 datadir point to woldan2 urls, this is the main problem for layers not working.

Please check for other references in the datadir as well and report here in the issue. We should be able to script fix it

[lpasquali]

@randomorder

• concerning tile layers I was able to easily import them and the were imported fine, with correct default vaules of woldan3 which has improved datadir fixes for 2.18.x (not present in woldan2)
• concerning styles there was a bunch of stiles (~10) with woldan3 reference easiliy updated with sed
• with sed I updated also csw catalogs pointing to woldan2 in any occurrence of coverage.xml

[randomorder]

no furthere references to wold2 in the datadir? if so we can move on with validation and closing the issue

[EddyCatt]

Waiting for the client to perform testing activities

[EddyCatt]

Some notes that could help importing geostories from one instance to another

Tentatives:

use django dumpdata command to dump the whole model structure for geoapps
if not working use the following steps:

• log into the DB machine (usually is something like db4geonode) of the source
• log in PSQL inside the postgres docker machine
• run the following queries to copy the data
  \copy (select * from geoapp_geostories_geostory) to /tmp/geostories.csv delimiter ';';
  \copy (select * from geoapps_geoapp) to /tmp/geoapp.csv delimiter ';';
  \copy (select * from geoapps_geoappdata) to /tmp/geoappdata.csv delimiter ';';
• move the 3 csv from the source docker container into the destination one
• log into the destination database machine
• log in PSQL inside the postgres machine
• run the following queries
  \copy geoapp_geostories_geostory from /tmp/geostories.csv delimiter ';';
  \copy geoapps_geoapp from /tmp/geoapp.csv delimiter ';';
  \copy geoapps_geoappdata from /tmp/geoappdata.csv delimiter ';';

[randomorder]

ny news @EddyCatt ?

[EddyCatt]

They're still testing Woldan3. They planned to close the testing in the next few days

[EddyCatt]

OAW is now setting up a proxy (https://hgis.woldan.oeaw.ac.at/ )
While doing it, they've figured out that a few references are still absolute and linking to http://woldan3.arz.oeaw.ac.at
They're asking if, and eventually how, all absolute references can be changed to relative ones.

In the meantime, we've found out that some references are still pointing to the old woldan2 dev server and they should be cleared once and for all

[randomorder]

I would close this issue and create two new separate ones @EddyCatt

In the meantime, we've found out that some references are still pointing to the old woldan2 dev server and they should be cleared once and for all

Can we have the name or links to those resources?