Open moula opened 1 month ago
Ah I forgot, the charmed-postgresql.pgbackrest-service never restarts, even manually.
Hi @moula
Thank you for giving it a try. I would like to ask you whether you set the max_connections to 170 based on README: https://github.com/canonical/maas-anvil?tab=readme-ov-file#postgresql-max_connections. Or else, what are the database options that you chose during bootstrap?
Hi @skatsaounis
No, I didn't put 170, I left "Default". If I have to start again with 170, I will do it again. I just entered my external network 192.168.70.0/24 so that it would be taken as my public network but it didn't change anything on the first node.
I've made progress since my last deployment. It's a little slow because I have 2 networks.
Public Network: 192.168.70.0 && Pxe-Network:172.200.0.0/24
Merci
Please try again with 170. As far as which network to choose for first question, please choose the one that your MAAS Anvil nodes can communicate each other so that Juju and micro-cluster that is used by MAAS Anvil can form a cluster.
I think that you will have positive results :slightly_smiling_face:
Please try again with 170. As far as which network to choose for first question, please choose the one that your MAAS Anvil nodes can communicate each other so that Juju and micro-cluster that is used by MAAS Anvil can form a cluster.
I think that you will have positive results 🙂
Ok, i will try.
hi @skatsaounis
I started everything again from the beginning. Everything is working...
A few comments for now.
There is slowness with postgresql and downloading OS images.
I get an error message: awaiting for member to start on node 2, even if I restart everything.
charmed-postgresql.pgbackrest-service still fails to start.
With this version I no longer have the bug in comminssioning physical machines like with the maas 3.6 version.
If you have any questions I am at your disposal and the rest will be for tomorrow. Thanks.
hi guys, It's a good project, in the same style of sunbeam-microstack. I'm testing, but it doesn't go all the way like the UI interface, but the rest works like high availability with VIP. I have a problem: on the first node the public network always gives me the IP of the second PXE network, even though I enter it when choosing the cidr, while the other 2 nodes give me the correct public IP. see image 1.![Juju-status](https://github.com/user-attachments/assets/77970b44-cb6c-4c90-b145-ac3ad83b7bea)
For Maas's UI configuration, I have an infinite loop between frame 1 and frame 2. Thank you very much for this good work, I will continue my tests.