John Spray

Results 144 issues of John Spray

There are various places where we have things like is_zero, is_unsharded. They're also all `pub` because of being in a different crate than the pageserver code. [ ] Make a...

c/storage/pageserver
a/tech_debt

See https://github.com/neondatabase/neon/issues/5507 for the related pageserver work. RFC in progress here: https://github.com/neondatabase/neon/pull/5432 ## Tasks - [ ] Update configuration to accept a list of pageserver connection strings, and metadata on...

t/feature
t/Epic
c/storage/compute

We use "keyspaces" in some places, which represent contiguous ranges of keys. In a single-sharded tenant, these contiguous ranges generally correspond to the ranges of used blocks within a relation....

c/storage/pageserver
a/tech_debt

Seen here: https://neon-github-public-dev.s3.amazonaws.com/reports/pr-6246/7385000585/index.html#suites/a45a54de0f437467df434b0529cc0b71/ed8d2b4496929aa5 This was a test run on a PR that only touches pageserver code, so should be unrelated to the failure. ``` 2024-01-02 11:13:34.878 GMT [32563] LOG: got...

t/bug
c/PostgreSQL
c/storage/safekeeper

The first phase of disk usage eviction is to enumerate all layers across all tenants, so that the layers can then be globally ordered by LRU. This generates an O(n_layers)...

c/storage/pageserver
a/tech_debt

Compaction runs every 20 seconds, heatmap uploads intentionally less frequent than that. The idea of heatmaps is to enable secondary tenants to prime their cache with long-existing large data, not...

c/storage/pageserver
a/tech_debt

in #6123, the `DOWNLOAD_FRESHEN_INTERVAL` constant controls how often we will poll remote storage for new content in a secondary location. The upload interval for heatmaps is configured externally. We could...

c/storage/pageserver
a/tech_debt

In #6123, periodic download of heatmaps is added for secondary mode locations. While downloading the heatmap is not bandwidth intensive, we can avoid transferring any body at all if the...

c/storage/pageserver
a/tech_debt

Two reasons: - For big puts, there isn't much value in running lots at the same time: assuming a 25gbps link and each flow achieving at least 100mb/s, you don't...

t/feature
c/storage/pageserver

This is a buggy alert: it alerts on cached memory low, assuming that means that the resident pageserver memory must be high. But of course the cached memory is also...

t/bug
c/storage/pageserver