Changeset View
Changeset View
Standalone View
Standalone View
data/hostname/vangogh.euwest.azure.internal.softwareheritage.org.yaml
- This file was added.
# open objstorage api | |||||
swh::deploy::objstorage::backend::listen::host: 0.0.0.0 | |||||
olasd: Is this really used? I think all "cache objstorage" requests are proxied through the vault rpc… | |||||
ardumontAuthorUnsubmitted Done Inline ActionsPossibly, i ported the existing orangerie setup. ardumont: Possibly, i ported the existing orangerie setup.
I'll check. | |||||
swh::deploy::vault::backend::listen::host: 0.0.0.0 | |||||
# plug it to the azure objstorage running on the same node | |||||
olasdUnsubmitted Done Inline ActionsDoes anything on that machine really use the swh::remote_service::vault variables? olasd: Does anything on that machine really use the `swh::remote_service::vault` variables? | |||||
ardumontAuthorUnsubmitted Done Inline Actionsi misunderstood its use ;) It will have to be defined in a next step in the defaults.yaml as the real switch between orangerie and this machine happens. ardumont: i misunderstood its use ;)
So to answer, no, nothing uses that.
I will remove it.
It will have… | |||||
swh::remote_service::vault::config: "%{alias('swh::remote_service::vault::config::azure')}" | |||||
swh::remote_service::vault::config::writable: "%{alias('swh::remote_service::vault::config::azure')}" | |||||
# vault's cache backend is an azure objstorage | |||||
swh::deploy::vault::config::cache: | |||||
cls: azure | |||||
args: | |||||
account_name: "%{hiera('swh::azure::credentials::vaultstorage::account')}" | |||||
api_secret_key: "%{hiera('swh::azure::credentials::vaultstorage::key')}" | |||||
container_name: contents | |||||
# still need access to the same objstorage | |||||
swh::deploy::objstorage::config: | |||||
objstorage: "%{alias('swh::deploy::vault::config::cache')}" | |||||
client_max_size: 1073741824 # 1 GiB | |||||
olasdUnsubmitted Not Done Inline ActionsCan be dropped if the objstorage rpc server deployment isn't needed. olasd: Can be dropped if the objstorage rpc server deployment isn't needed. | |||||
swh::remote_service::storage::config: "%{alias('swh::remote_service::storage::config::azure')}" | |||||
# No workers | |||||
swh::deploy::worker::instances: [] | |||||
olasdUnsubmitted Not Done Inline ActionsNot sure that's needed? olasd: Not sure that's needed? | |||||
ardumontAuthorUnsubmitted Done Inline ActionsIt's because that node runs on azure. ardumont: It's because that node runs on azure.
By default there, we have lots of indexer workers running. |
Is this really used? I think all "cache objstorage" requests are proxied through the vault rpc server.
I don't think we need the objstorage rpc server deployed at all on this machine, actually, as your previous change makes the deployed vault backend use the objstorage backend config directly.