OS-OTT Xen blkfront/tapdisk stall evidence
Incident date: 2026-07-30 (MSK, UTC+03:00)
Affected VM: os-ott-data-3-2 / 172.30.52.193
Hypervisor: 172.30.50.191

Timeline:
- 00:52:21: OpenSearch controller detected the missing data node.
- 00:52:54: full guest/backend stall signature confirmed.
- 00:52:54-00:52:56: automatic GDB capture completed.
- 01:00:23: additional guest and hypervisor snapshot collected.
- 01:02:56: diagnostic window expired; revalidation started.
- 01:03:25: signature confirmed again and forced VM reboot requested.
- 01:07:08: node rejoined the OpenSearch cluster.
- 01:59:56: cluster returned to green.

Key distinction from the 2026-07-25 occurrence:
- max_ring_page_order=3 was active and eight ring pages were negotiated.
- Guest: 255 requests in flight and all 256 tags busy.
- Backend: empty ring, tapdisk request counters 0/0, reqs_outstanding=0.

Files:
- 20260730-005220-gdb-172.30.52.193.txt
  Complete pre-reboot GDB and tap-ctl capture.
- 20260730-pre-reboot-guest-172.30.52.193.txt
  Guest queue settings, blk-mq state, diskstats samples, PSI, D-state stacks
  and kernel journal.
- 20260730-pre-reboot-hypervisor-172.30.50.191.txt
  Sanitized tapdisk/xenstore state and daemon.log/SMlog incident excerpt.
- SHA256SUMS.txt
  SHA-256 hashes of the evidence files.