Data Platform Engineering/Data Platform SRE/Status Update/2024-12-13

Airflow migration to k8s

• T381264 Connectivity issues with the ceph CSI provisioner pods

WDQS Graph Split

• T379330 Create pybal pools for wdqs-internal-main and wdqs-internal-scholarly (internal SPARQL endpoints for the graph split are ready, we're waiting until January to reconfigure the clients to use them)

DB replica

• T379724 s8 replication on an-redacteddb1001 is broken (fixed by our DBA team)
• T376726 Hide autoblocks from the globalblocks table database dump

Graphite deprecation

• T371061 Update CirrusSearch dashboards to use new metrics/refresh dashboards

Hardware

• T378030 Q2:rack/setup/install wdqs102[567] (new servers racked and with initial install, but not active yet)
• T376166 Q2:eqiad:(2)cloudelastic hosts - custom config (new servers received by DCOps)
• T376165 Q2:codfw:(6) elastic hosts - Config D (new servers received by DCOps)
• T376670 Q2:eqiad:(3)Refresh of wdqs101[1-3] (new servers received by DCOps)

Misc / Operations

• T377134 Create and distribute a flink base image with flink 1.20.0
• T381283 wdqs1025 fails to PXE boot, NIC shows "no link" in DRAC web UI
• T380258 Create an Airflow instance for ML (The work on the migration to k8s is paying of already! After some time to define the exact requirements, it took less than a day to create that new instance!)
• T381961 Increase the capacity of /var/lib/archiva on archiva1002.wikimedia.org
• T377266 DSE kubernetes namespace for llm-inference
• T380835 Exclude zram devices from disk health checks