Explore

@netappblackbox.bsky.social Sep 16, 2026, 6:04 PM

A background disk scrub can stop your upgrade dead.

ONTAP vetoes takeover while a scrub, reconstruction or sanitization is still running. Your ANDU just sits there.

raid.rg.scrub.summary is a NOTICE, not an error. Read it before you book the window, not during it.

#ONTAP

EMS OF THE DAY card, severity NOTICE. Event: raid.rg.scrub.summary. What it means: a background RAID-group scrub finished and logged what it found, how much it read and the media errors it corrected. Routine work, until you schedule an upgrade, because a scrub or reconstruction still in flight will veto takeover. Check first: event log show -node node1 -event raid.* -time >24h gives the summary with its media-error count; storage disk show -reconstruction shows whether a rebuild is still in flight on any aggregate; storage failover show-takeover -fields node,is-takeover-possible,reason names what is vetoing takeover. Not the cause: a scrub summary is not a data-loss report, RAID rebuilt the block and logged it, a dead disk shows up as broken; and a scrub is low-priority background work, not your latency problem. NETAPP BLACK BOX // netapp-blackbox.pages.dev

@netappblackbox.bsky.social Sep 16, 2026, 12:02 PM

A volume can be 12% full and still refuse to create a file.

inodes are fixed per volume. Files, not bytes; capacity alarms never see it.

Full notes: https://netapp-blackbox.pages.dev/troubleshooting/volume-space?utm_source=bluesky&utm_medium=social&utm_campaign=w4&utm_content=inodes
#ONTAP

Quick reference card, THE DOCS DON'T SAY THIS. An inode is one entry per file, and ONTAP fixes how many a volume can hold, so a volume can be 12 percent full and still refuse to create a file. Commands and what each answers: volume show -fields files,files-used gives inodes used versus the volume max; volume show -fields percent-inode-used gives how close the volume is to its file ceiling; volume file show-dir-usage lists the directories holding the most files; df -i shows client-side inode use; find -xdev -type f piped to wc -l counts files per directory on the client; volume modify -files raises the ceiling, capped by the volume max. Free gigabytes are bytes, the inode table is files. NETAPP BLACK BOX // netapp-blackbox.pages.dev

@netappblackbox.bsky.social Sep 16, 2026, 8:03 AM

Failover moves the IP. It does not move the session.

Hard mount: the app stalls, then resumes.
Soft mount: EIO, and the app picks up the pieces.
Failback disabled: you stay on the partner until someone notices.

Small print: soft mounts are usually the bug, not the failover.
#ONTAP

Myth and reality card. Myth: LIF failover is invisible to NFS clients - 'NFS is stateless' is a protocol argument, not an operational one. Reality: failover moves the address, not the session. A hard mount stalls the application and resumes where it left off; a soft mount returns EIO and the application decides what to do with it; with failback disabled the LIF stays on the partner until an admin notices. Ask what the client was told to wait for: hard or soft, one LIF or two. NETAPP BLACK BOX

@netappblackbox.bsky.social Sep 15, 2026, 8:09 PM

ONTAP logged wafl.vol.outOfInodes on a volume with 400 GB free.

The alarm is not about space.

What is actually exhausted?

A. the aggregate
B. the inode table
C. the Snapshot reserve

Explanation tomorrow.
#ONTAP

Question card: ONTAP logged wafl.vol.outOfInodes on a volume that still has 400 gigabytes free. Options: A the aggregate is out of space, B the inode table (file count, not bytes), C the Snapshot reserve. What is actually exhausted?

@netappblackbox.bsky.social Sep 15, 2026, 8:08 PM

Most ONTAP arguments are two people describing different rungs of the same ladder.

SVM > LIF > volume > protocol > host.

Debug the rung that owns the symptom.

https://netapp-blackbox.pages.dev/reference/glossary?utm_source=bluesky&utm_medium=social&utm_campaign=w4&utm_content=ontap-ladder
#ONTAP

Storage map: ONTAP object ladder from SVM to host. SVM is the tenant boundary with its own addresses and credentials; a LIF is the only address a client ever sees and the unit that fails over; a volume is the data boundary with its own Snapshot policy and junction; the protocol (NFS, SMB, S3, iSCSI) is how it is exposed; the host is the consumer. Debug the rung that owns the symptom.

@nwexam.bsky.social Sep 14, 2026, 11:05 AM

Is NCDA Certification worth it?

Before you decide, explore the NCDA certification cost, requirements, training, and salary considerations.

🔗 www.nwexam.com/blog/you-dec...

#NCDACertification #NetApp #ONTAP #TechCareer

@netappblackbox.bsky.social Sep 5, 2026, 10:28 AM

A noisy-neighbor VM pegging the array isn't a capacity problem, it's a policy problem. Adaptive per-workload QoS throttles the spike, so one loud workload can't force a capacity buy. Policy beats over-provisioning. https://netapp-blackbox.pages.dev #NetApp #ONTAP

@netappblackbox.bsky.social Sep 3, 2026, 2:28 PM

NVMe-oF/TCP is ONTAP’s sleeper feature: NVMe queues over ordinary 25/100GbE — no FC fabric, no RoCE tuning. vSphere 8 and current Linux speak it natively, so on an AFF like this one “NVMe everywhere” becomes a cable decision, not a rip-and-replace. #NetApp #ONTAP #NVMe

@netappblackbox.bsky.social Sep 3, 2026, 11:19 AM

FabricPool demotes cold ONTAP blocks to S3/StorageGRID after a cutoff (default 31 days). Two things people miss: set the cutoff per volume, and remember the object tier is not a backup — snapshots and SnapMirror run on the active tier. https://netapp-blackbox.pages.dev/reference/ #NetApp #ONTAP

@netappblackbox.bsky.social Sep 3, 2026, 4:59 AM

A SnapMirror relationship that’s “healthy” but perpetually lagging is a data-loss window wearing a backup’s badge. Alert on lag-time, not just last transfer time — details in the DR reference: https://netapp-blackbox.pages.dev/hubs/disaster-recovery.html #NetApp #ONTAP #Backup