Certain endpoints in SS are expected to return immediately (e.g. get_pipeline) but we have a single deadline that applies to all the requests and it's normally set very high (1 day) to deal with some long-running requests that SS can't delay (STORAGE_SERVICE_CLIENT_TIMEOUT=86400).
This pull request only attempts to create a new timeout for those endpoints that are supposed to be fast (e.g. `get_pipeline) so when the SS is unreachable we have a short deadline instead of the recently added deadline meant to be used for long-running operations (which should be delayed anyways but that's being solved separately, i.e. ask Justin about his plans for async ops in SS).
New environments:
In Dashboard: ARCHIVEMATICA_DASHBOARD_DASHBOARD_STORAGE_SERVICE_CLIENT_QUERY_TIMEOUT=5
In MCPClient: ARCHIVEMATICA_MCPCLIENT_MCPCLIENT_STORAGE_SERVICE_CLIENT_QUERY_TIMEOUT=5
I couldn't fin da better name. STORAGE_SERVICE_CLIENT_QUERY_TIMEOUT=5 kind of made sense because we already have STORAGE_SERVICE_CLIENT_TIMEOUT=86400 for slow requests. I'm open to suggestions.
RDSSARK-425
Certain endpoints in SS are expected to return immediately (e.g.
get_pipeline
) but we have a single deadline that applies to all the requests and it's normally set very high (1 day) to deal with some long-running requests that SS can't delay (STORAGE_SERVICE_CLIENT_TIMEOUT=86400).This pull request only attempts to create a new timeout for those endpoints that are supposed to be fast (e.g. `get_pipeline) so when the SS is unreachable we have a short deadline instead of the recently added deadline meant to be used for long-running operations (which should be delayed anyways but that's being solved separately, i.e. ask Justin about his plans for async ops in SS).
New environments:
ARCHIVEMATICA_DASHBOARD_DASHBOARD_STORAGE_SERVICE_CLIENT_QUERY_TIMEOUT=5
ARCHIVEMATICA_MCPCLIENT_MCPCLIENT_STORAGE_SERVICE_CLIENT_QUERY_TIMEOUT=5
I couldn't fin da better name.
STORAGE_SERVICE_CLIENT_QUERY_TIMEOUT=5
kind of made sense because we already haveSTORAGE_SERVICE_CLIENT_TIMEOUT=86400
for slow requests. I'm open to suggestions.