There was a bundle generated with broken files (#238).
This is what (#211) is intended to prevent, but wasn't/isn't implemented yet.
I thought about making this part of (#236) but it seems like a large enough thing to do on it's own.
In essence, bring every single bundle back and make sure all the checksums match.
Note any bundles that are broken, and regenerate them (if possible).
When the new k8s based pipelines are redeployed, this should also be the idle task for the pipeline;
if we're not transferring something else, we should pull bundles from tape to check that they're still good.
There was a bundle generated with broken files (#238). This is what (#211) is intended to prevent, but wasn't/isn't implemented yet.
I thought about making this part of (#236) but it seems like a large enough thing to do on it's own.
In essence, bring every single bundle back and make sure all the checksums match. Note any bundles that are broken, and regenerate them (if possible).
When the new k8s based pipelines are redeployed, this should also be the idle task for the pipeline; if we're not transferring something else, we should pull bundles from tape to check that they're still good.