JuJing OBSERVE v2.4.0: Log Clustering, Slow-Query Diagnosis, and Alert Templates

Highlights

v2.4.0 targets two recurring pain points: log volume that's too large to read, and slow database queries that are hard to pinpoint.

Log clustering: the search results page gains a "Clustering" view that auto-groups massive logs by similarity, showing a sample, count, and time range for each group. During a log storm, read the clusters first — it's an order of magnitude faster than skimming line by line. Clustering is based on token similarity, not keywords, so Connection reset by peer: 10.0.1.23 and 10.0.1.45 land in the same group.

SQL slow-query diagnosis: the collector now captures queries exceeding a threshold and correlates execution plans with index usage. The diagnosis panel reports three things: whether an index was used, rows scanned, and suggested missing indexes.

Alert-rule template library: 40+ built-in templates covering hosts, containers, middleware, and applications. Pick one, tweak the threshold, done — no hand-written PromQL.

Other changes

• The search engine reworked part of its index-merge logic; aggregate queries are ~35% faster on average.
• Alert notifications now support WeCom, DingTalk, and Feishu, with per-severity routing (P0 = phone + SMS, P1 = IM only).
• Dashboard variables support cascading — switching environments auto-fills the matching service list.
• Fixed a timezone-offset bug during daylight-saving transitions.

Upgrade steps

Back up the metadata database before upgrading, then follow this order:

# 1. Back up
jjhub-backup --type meta --output /backup/meta-v2.3.1.tar.gz

# 2. Stop the old version
systemctl stop jjhub-server jjhub-collector

# 3. Install the new package
rpm -Uvh jjhub-server-2.4.0.el7.x86_64.rpm

# 4. Run the migration (automatic, ~2 minutes)
jjhub-migrate --from 2.3.1 --to 2.4.0

# 5. Start and verify
systemctl start jjhub-server jjhub-collector
jjhub-health-check

For rolling upgrades, upgrade the Collector before the Server; they're forward-compatible by one version. A v2.4.0 Collector reports fine to a v2.3.1 Server, and vice versa.

Compatibility and notes

• Minimum supported version: you can upgrade directly from v2.3.0; earlier versions must first go to v2.3.1.
• Slow-query collection is off by default. Set slow_query.enabled=true in the access config to enable it; otherwise there's no added overhead.
• Log-clustering indexes consume roughly 15% more storage; you can disable it per namespace under "Storage Settings".
• Known issue: the clustering view may load slowly beyond 5 million lines; a fix lands in the next minor release.

See the docs center for the full changelog.