| 14 Jan 2025 |
 Andy | I understand now. Thanks Alexander! | 15:59:20 |
| 19 Jan 2025 |
 krabina | Another semantic wiki has died: https://github.com/AKSW/OntoWiki/issues/441
OntoWiki (with the last commit from 2017). Since 2024 it is officially archived on GitHub.
| 12:28:04 |
 hexmode | looks like it was gone since 2021, though. And they didn't want to do the work to go from php 5 to php 7 | 15:23:18 |
| 28 Jan 2025 |
 freephile | Anyone want to comment on this PR?
It is documentation-only (tests/README). I think it's good to merge. It basically gets rid of the deprecated TestCase syntax and improves the video link(s) for 'how to test' plus fixes a couple grammar points.
| 14:57:18 |
| 1 Feb 2025 |
|  @bulder:matrix.org joined the room. | 20:05:15 |
|  Romi Jokela joined the room. | 20:06:17 |
| @bulder:matrix.org left the room. | 20:10:44 |
| krabina | Did anyone ever set properties via a template?
Like this? on Property:Endpoint URL
{{Property
|Imported from=dcat:endpointURL
|Title=Endpoint URL
|Description=....
| 21:35:00 |
| krabina | with the template Property doing the actual annotations that are usually on the property page directly= | 21:35:38 |
| krabina | * with the template Property doing the actual annotations that are usually on the property page directly? | 21:35:46 |
 revansx | We do that all the time. | 22:16:50 |
| revansx | anything we should look out for? | 22:17:32 |
| krabina | I have one wiki where I did did this for the first time. I have some strange behaviour. Sometimes the property do not get the right datat type. I am using "imported from" declarations a lot. Sometimes the properties seam to los their datatype that is declared in the vocabulary definition and the only thing that helps is to null edit the property page. | 23:21:06 |
| krabina | * I have one wiki where I did did this for the first time. I have some strange behaviour. Sometimes the property do not get the right datat type. I am using "imported from" declarations a lot. Sometimes the properties seem to los their datatype that is declared in the vocabulary definition and the only thing that helps is to null edit the property page. | 23:21:21 |
| krabina | * I have one wiki where I did did this for the first time. I have some strange behaviour. Sometimes the property do not get the right datat type. I am using "imported from" declarations a lot. Sometimes the properties seem to lose their datatype that is declared in the vocabulary definition and the only thing that helps is to null edit the property page. | 23:21:30 |
| 2 Feb 2025 |
| revansx | We hired Yaron to fix behavior like that and he came up with a simple solution that he submitted a patch for. Until his patch gets implemented we have to maintain our SMW with it as a local hack | 02:40:48 |
| krabina | Is there a PR with the patch anywhere? | 08:00:19 |
| revansx | https://github.com/SemanticMediaWiki/SemanticMediaWiki/pull/5741 | 15:11:54 |
| 12 Feb 2025 |
|  Paladox joined the room. | 19:40:50 |
| Paladox | Hi, can anyone tell me if https://github.com/SemanticMediaWiki/SemanticMediaWiki/issues/5929 is a bug or if I configured our changeprop wrongly? | 19:41:31 |
| Paladox | I'm thinking the former. | 19:41:38 |
| 13 Feb 2025 |
| freephile | I don't think this is a bug in SMW but rather a configuration / responsibility of your Gateway/Proxy (HAProxy).
You need to configure HAProxy to work with the job dispatcher in a way that ensures jobs are dispatched consistently to the same backend using so-called "sticky sessions" or consistent hashing. This will help ensure that related jobs are sent to the same backend server, which can be crucial for maintaining state and consistency. | 15:21:22 |
| Paladox | rundg: Oh, do you know how I would do this? I created a basic haproxy config to load balance the jobrunners as the dns wasn't very efficient. My config is: https://github.com/miraheze/puppet/blob/master/modules/role/templates/jobrunner_haproxy/lb.cfg.erb | 15:37:56 |
| Paladox | (saying that I think wikimedia use the dns to load balance? or rather lvs. I'm not sure how they handle sticky sessions.) | 15:38:11 |
| freephile | yep, there are quick hits via Google for sticky sessions in HAProxy. Notice the ACLs.
Here is an example configuration using sticky sessions based on a cookie:
frontend http_front
bind *:80
acl url_static path_beg -i /static /images /javascript
acl url_static path_end -i .jpg .gif .png .css .js
use_backend static if url_static
default_backend dynamic
backend static
balance roundrobin
server static1 192.168.1.50:80 check
server static2 192.168.1.51:80 check
backend dynamic
balance roundrobin
cookie SERVERID insert indirect nocache
server app1 192.168.1.10:80 check cookie app1
server app2 192.168.1.11:80 check cookie app2
| 15:43:27 |
| Paladox | Oh thanks! Will the cookie even work since it's browserless (changeprop is triggering it) | 15:44:31 |
| Paladox | I'll give that a go I guess | 15:44:38 |
| freephile | (the acl is just to divert static traffic away from your 'backends') | 15:44:45 |
| freephile | I'm not sure about the internals of eventgate, changeprop, kafka and how the requests are handled in the big picture. I'm just thinking that if they come through HAProxy, that's where you can establish the persistence. | 15:45:58 |
| Paladox | I think https://github.com/miraheze/puppet/pull/4186 will work | 15:49:51 |
