DM-45494: Add fastapi-safir-uws starter #3573
Conversation
|
This depends on lsst-sqre/safir#281 and lsst/templates#260. |
| cpu: "5m" | ||
| memory: "7Mi" | ||
|
|
||
| cutoutWorker: |
There was a problem hiding this comment.
Seems like this should be generic, like "worker" or "pipelinesWorker" ?
There was a problem hiding this comment.
Thank you, good catch. I grepped for cutouts and not cutout. Found and fixed all of these, I think.
| enabled: true | ||
|
|
||
| # -- Amount of persistent storage to request | ||
| size: "100Mi" |
There was a problem hiding this comment.
I wonder if it's better to error on the side of being a bit larger since resizing PVCs is a pain. Up to you, though.
There was a problem hiding this comment.
I was going to say that the amount of data stored should be quite small since the largest thing stored here in this pattern is the arguments to the backend, but the vo-cutouts data is about 50MiB, so maybe it will be larger than I expect? Although that's also very surprising to me; that's way more data than I was expecting. I'll dig into this a bit.
There was a problem hiding this comment.
Aren't task outputs also stored here, including exception traces?
There was a problem hiding this comment.
Outputs are not, only a URL. Exception traces are, but only for a day.
It looks like Redis is not periodically cleaning up the appendonly database for some reason. I may be missing some command-line flags in our Redis chart. When I manually ran BGREWRITEAOF, the size of the vo-cutouts Redis database went down to about 1KB.
Add a starter for vo-cutouts-style FastAPI Safir apps that use UWS. Update the code underlying phalanx application create so that it will substitute <CHARTENVPREFIX> in every template, not just configmap.yaml.
Add a starter for vo-cutouts-style FastAPI Safir apps that use UWS. Update the code underlying phalanx application create so that it will substitute
<CHARTENVPREFIX>in every template, not justconfigmap.yaml.