We've packaged 43 scripts from our years as senior DBAs and made them publicly available. Use them whether or not you are a SentinelDB360 customer — graduate to the platform when your needs grow.
There are plenty of repos dumping piles of DMV trivia. The DMC DBA Toolkit speaks differently: it's what a senior DBA says the tenth time they see the same problem. Every script answers one question — "when do I actually run this, and why?"
Every .sql file is split: English at the top — "When this happens, run this first..." — Turkish below. DBA expertise is language-agnostic; we don't gate it.
Doesn't read like an AI dump because it isn't one. Comments aren't "column X returns Y" — they're "if X > 100, worry." Each script ships with its own reading guide.
Top wait, top resource-consuming queries, blocking chain hunter, plan cache analysis, fragmentation report, missing index advisor — the 12 most-checked scripts in the field.
pg_stat_statements top, replication slot lag, bloat hunter, index usage stats, vacuum/autovacuum diagnostic, lock tree, transaction wraparound monitor.
performance_schema top queries, InnoDB buffer pool stats, replication lag, slow query digest, lock waits, undo log growth — daily DBA work, packaged.
currentOp long-runners, replication lag, oplog window, sharding chunk distribution, index hit ratio, collection stats, profiler summary.
"Customer is on the phone, system is slow, who is blocking whom?" — A single query that prints the blocking tree with real names and shows the head blocker query text.
"There are 200K plans in cache, how many ran only once?" — Single-use plan ratio, parameterization status, "optimize for ad hoc workloads" recommendation.
Not "did VACUUM run?" — but "which table is genuinely bloated, how many GB are wasted, what would REINDEX cost?" — produced as an action list.
Three columns answer "how far behind is the standby?": byte lag, time lag, slot age. If a slot is orphaned, you get a "WAL piling up, disk will fill" warning.
Hit ratio, dirty page ratio, free page count in one view. Comments include an upper-bound recommendation for "is the buffer pool too small?"
"What in currentOp has been running over 5 minutes?" — flags COLLSCAN instead of sorting, suggests an index advisor, and offers a one-line kill command.
The toolkit is run by hand — a great start, but not sustainable monitoring. Once you have 5+ sources or need 24/7 oversight, the platform becomes the right answer. Both come from the same philosophy and the same team.
1–3 servers, manual review, single-DBA workflow. Open source, free, clone from GitHub, run.
5+ servers, 24/7 monitoring, AI insights, alerts, compliance, team dashboards. The knowledge inside the toolkit is automated in the platform.
DBA Toolkit is MIT-licensed, CI is green, and a contribution guide is ready. Add your own senior expertise as a script — follow the bilingual EN+TR convention and we'll review it.
Commercial use, modification, distribution — all allowed. The only requirement: keep the license/copyright notice.
Every PR runs linting + syntax checks + a smoke test on a sample docker DB.
v2.1.0 active. CHANGELOG.md is maintained, breaking changes get a major bump.
Toolkit content is curated under MVP guidance, drawn from field-tested scripts.
Start using the Toolkit today; the platform is here when you're ready.
