Reported issue: Memory overload when connected to RBN

[AI3I]

Hi gentlemen,

My colleague R1BLH who uses pyCluster said that it goes "crazy" when connection to RBN is established.
It's freezing , and memory it seems overloading.
Just for information.

---

73 Anton RC3C & RY3AAL
Node RC3C-2 / dxc.rc3c.ru:17300
http://dxc.rc3c.ru:11111/

[AI3I]

Investigated the vague RBN freeze/memory report. I found two plausible pressure points in direct RBN ingest: every accepted RBN line was committed to SQLite individually, and live RBN display aggregation could grow without a hard cap while a high-volume feed was active. Implemented local mitigations: direct RBN feed ingest now batches DB inserts and flushes on idle/EOF, yields during burst fanout, and telnet live RBN grouping now has a bounded queue that prunes dead-session/old groups first. Added regression coverage for duplicate-aware batch insert, existing direct RBN feed ingest/filter/multi-feed behavior, and the live RBN group cap. Verified with: pytest tests/test_store.py tests/test_telnet_commands.py tests/test_app_ingest.py tests/test_web_admin.py (396 passed).

[AI3I]

Committed in 3907f4e and deployed to pycluster.ai3i.net. Direct RBN ingestion now batches accepted spot storage, yields during burst fanout, and caps live RBN aggregation groups to limit memory growth/freezing under busy feeds. Covered by store/app/telnet regression tests plus deployed-host targeted tests. This should still get a longer RBN soak test from an affected node/operator before we call the field report fully closed.