Crawler health check is a little bit too zealous
exarkun opened this issue · 1 comments
exarkun commented
The health check introduced in #687 is signaling unhealth for the storageserver container more often than necessary. This is resulting in the storageserver being restarted more often than necessary.
This isn't catastrophic since the storageserver comes right back up. It probably leads to more S3 API calls than necessary since we lose the accounting state after every storageserver restart. Figure out why the liveness probe is signaling unhealth and fix it.