Adaptec Series 7 RAID Controller Data Recovery: ASR-7805, 71605, 72405 Error Messages and Failure States

If an Adaptec Series 7 controller — the ASR-7805, 7805Q, 71605, 71605Q, 71605E, 71685, 72405, or 78165 — is showing a degraded, impacted, failed, or missing array, this page is written for the next ten minutes of your day. Series 7 cards were introduced in 2012 as the first Adaptec generation designed after the PMC-Sierra acquisition, and the last public firmware for the family dates to 2018. Thousands of them are still running file servers, video storage, and virtualization hosts in small businesses that have never had a reason to replace them — until now. Below is every message we could document from Adaptec’s manuals, knowledge base, the Linux kernel source, Windows event logs, and years of administrator forum threads, each rated by how much danger your data is in and which keystroke turns a recoverable array into an unrecoverable one.

Gillware has been recovering failed RAID arrays for more than two decades from our lab in Madison, Wisconsin. If your Series 7 array is already offline and the data matters, you can skip the reading: power the server down and book a free consultation with one of our enterprise recovery experts.

Before You Press Anything: The Series 7 Actions That Destroy Arrays

Most of the Series 7 arrays that reach a recovery lab were recoverable the moment they failed. What made them hard was a well-intentioned action taken afterward. Every one of the operations below writes to the drives, and several of them appear in Adaptec’s own troubleshooting articles as legitimate next steps. They are legitimate — when the array’s history is simple and every member is healthy. When it isn’t, each is a one-way door.

• Initializing a drive in the Ctrl-A utility (Initialize Drives) or with ARCCONF TASK START ... INITIALIZE. arcconf warns that initializing a device “will erase its metadata, the section where all the logical device definition data is stored.” That metadata is the map of your array.
• Uninitializing a drive (Uninitialize in the BIOS utility, ARCCONF UNINIT). Same effect by a different name — and on a controller in the default RAID: Expose RAW mode, an uninitialized drive is immediately handed to the operating system as a raw disk.
• Forcing a failed array online (Ctrl+F in the BIOS utility, Force Online in maxView, or ARCCONF SETSTATE <Controller#> LOGICALDRIVE <LD#> OPTIMAL [ADVANCED]). If a stale member is folded back in, the controller will treat out-of-date data as current and may start repairing good parity with bad data.
• Recreating the array “without initialization.” Adaptec documents a skip-initialize re-creation as a recovery technique. It only works if every parameter — RAID level, stripe size, member order, member set, size — matches the original exactly, and it writes brand-new metadata over the old metadata the moment you commit.
• Changing the controller mode to HBA, Auto Volume, or Simple Volume with array members attached. Adaptec’s own guidance is that HBA mode should only be selected when no drives carrying Adaptec metadata are connected.
• Clearing preserved cache (ARCCONF PRESERVECACHE ... CLEAR or the equivalent BIOS option). Preserved cache is write data the controller held back because the target array was unavailable. Clearing it discards those writes permanently.
• Resetting the configuration (ARCCONF SETCONFIG <Controller#> DEFAULT). Adaptec’s CLI documentation describes this as deleting logical drives and resetting hard disks to the Ready state.
• Clicking “Initialize Disk” in Windows Disk Management when an unfamiliar disk appears after an array failure. On a Series 7 card, a member that has lost its Adaptec metadata can surface to Windows as a raw disk. Initializing it writes a new partition table to that member.
• Moving the drives to a different controller generation. A Series 8 card does not import Series 7 arrays. The drives will be listed as Ready, with no foreign-configuration option offered — and any configuration action taken from that point writes to them.
• Starting or restarting a rebuild on a degraded array whose surviving members are throwing errors. A rebuild reads every sector of every surviving drive; that is exactly the workload that pushes a marginal second drive over the edge.

If you have already done one of these things, stop there. Do not attempt a second one to undo the first. Most of these operations are recoverable from a lab’s perspective if nothing else is written after them.

The Adaptec Series 7 Family at a Glance

Every Series 7 card is built on PMC-Sierra’s 24-port PM8015 RAID-on-chip, with a PCIe 3.0 x8 host interface, 6 Gb/s SAS/SATA ports on mini-SAS HD (SFF-8643 internal, SFF-8644 external) connectors, and support for up to 256 drives through expanders. The models differ in port count, cache, and cache protection — and one of them, the 71605E, is a stripped-down card that cannot build parity arrays at all.

Model	Ports	Cache	Flash cache protection (ZMCP)	Form factor	Notes
ASR-7805	8 internal	1 GB DDR3	AFM-700 optional	Low-profile MD2	Most common 8-port Series 7 card
ASR-7805Q	8 internal	1 GB DDR3	AFM-700 pre-installed	Low-profile MD2	maxCache 3.0 SSD caching (up to 8 SSDs)
ASR-71605	16 internal	1 GB DDR3	AFM-700 optional	Low-profile MD2	The most widely deployed Series 7 card
ASR-71605Q	16 internal	1 GB DDR3	AFM-700 pre-installed	Low-profile MD2	maxCache 3.0
ASR-71605E	16 internal	256 MB DDR3	None	Low-profile MD2	Entry model: RAID 0, 1, 1E, and 10 only
ASR-71685	16 internal + 8 external	1 GB DDR3	AFM-700 optional	Full height, half length	Common with external JBOD shelves
ASR-72405	24 internal	1 GB DDR3	AFM-700 optional	Full height, half length	24-bay chassis without an expander backplane
ASR-78165	8 internal + 16 external	1 GB DDR3	AFM-700 optional	Low-profile MD2	External-storage focused

Supported RAID levels on the full cards are 0, 1, 1E, 5, 6, 10, 50, 60, plus Simple Volumes. RAID 5EE — Adaptec’s RAID 5 with a distributed spare — was a Series 6 feature and is not supported for new arrays on Series 7. Adaptec’s operating specification is 0–55 °C with 200 linear feet per minute of airflow, dropping to 0–50 °C with the flash module installed. That airflow requirement matters more than most owners realize; see the ASR-71605 page for what happens when a Series 7 card cooks in a tower case.

How Series 7 Controllers Store Your Array (and Why It Matters When Something Breaks)

The configuration lives on the drives

Series 7 cards run Adaptec RAID Code (ARC) firmware. The array definition — which drives belong to which array, in what order, with what stripe size and RAID level — is written to a reserved metadata area on every member drive. That is why a healthy Series 7 array can usually be moved to another Series 7 card with compatible firmware, and why the controller itself is rarely the thing that holds your data hostage. It is also why every operation that “initializes,” “uninitializes,” or “clears” a drive is so consequential: those operations erase the only map of the array that exists.

The same reserved area holds a Bad Stripe Table for each logical drive. When the controller cannot reconstruct a stripe — typically because a rebuild hit an unreadable sector on a surviving member, or a verify-with-fix found errors at the same location on multiple members — it records that stripe as bad and blocks it rather than returning corrupt data. Adaptec’s documentation is explicit that there is no procedure for clearing bad stripes while keeping the existing array.

Controller modes: the setting nobody remembers changing

Series 7 firmware added selectable controller modes, set in the Ctrl-A utility, the UEFI/HII setup, or with ARCCONF SETCONTROLLERMODE. Knowing which mode your card is in explains a lot of confusing symptoms:

• RAID: Expose RAW (the default). All RAID functions are enabled. Drives that are uninitialized or have no Adaptec metadata are passed straight through to the operating system as raw devices. This is the mode in which a member drive that loses its metadata suddenly appears in Windows Disk Management as an unknown disk.
• RAID: Hide RAW. RAID enabled; raw drives are not shown to the operating system. (Both RAW variants require firmware build 30855 or later.)
• Auto Volume. RAID enabled, but drives with no Adaptec metadata and no OS partitions are automatically converted into Simple Volumes — which means the controller writes its metadata to them.
• HBA. Every drive is passed through as a raw device. Adaptec’s guidance is to switch into HBA mode only when no drives with Adaptec metadata are attached.
• Simple Volume. RAID functions disabled; only single-drive Simple Volumes (up to 128) can be created. Requires firmware build 32033 or later.

Mode confusion is behind a recurring pattern in home-lab and small-business forums: a used 71605 is installed, existing data drives are connected to it, and the drives are unmountable after the first boot because the card was not in the mode its owner assumed.

Compatibility boundaries

• Series 7 to Series 7: a replacement Series 7 card with compatible firmware will normally recognize arrays created on another Series 7 card.
• Older Adaptec generations to Series 7: Adaptec’s compatibility guidance lists Series 7 as able to import arrays from its earlier AMF-metadata controllers, with the exception of RAID 5EE, which Series 7 does not support.
• Series 7 to Series 8 or SmartRAID: no import path. Drives are recognized but reported as Ready rather than Foreign, and no scan-for-foreign-configuration option exists. We cover this boundary in detail on our Adaptec controller page.
• Adaptec to any other vendor: no import path in either direction. LSI/Broadcom, Dell PERC, and HPE Smart Array firmware cannot read Adaptec metadata.
• AFM-700 flash modules: Adaptec’s release notes state that moving the AFM-700 daughterboard and supercapacitor from a Series 7 card to a Series 8 card (or the reverse) is not supported.

How to Read the Message Catalog

Each message below is tagged with one of four risk levels. The level describes the state of your data at the moment the message appears, not how alarming the wording sounds.

Risk level	What it means	What to do
Level 1 — Informational	No data at risk. Often benign or a side effect of maintenance.	Note it, confirm the array state, move on.
Level 2 — At risk	The array is running without its full redundancy, or a component that protects data (cache backup, cooling) has failed.	Back up now, before anything else. Avoid rebuilds until the backup is done if surviving drives show errors.
Level 3 — Data offline	The array or volume is not accessible. The data is usually still on the drives.	Power down. Do not run repair, rebuild, or configuration actions.
Level 4 — Destructive prompt	The screen is asking you to perform an operation that writes to the drives.	Do not confirm until you are certain — or until the drives have been imaged.

Messages at Boot (Adaptec BIOS / Option ROM)

“Controller detected configuration change” — “The following arrays have missing members and cannot be configured”

Risk: Level 3 (data offline) — with a Level 4 prompt attached.

Adaptec documents a family of POST messages that appear when the drives the controller expects are not the drives it finds. Depending on firmware generation, the wording includes Controller detected configuration change, New devices detected at the following location, a list of ports that “are not responding,” The following arrays have missing members and cannot be configured, or One or more drives are either missing or not responding. Please check if the drives are connected and powered on. Each is followed by the same three options: press Enter to accept the current configuration, press Ctrl-A to enter the configuration utility, or press Ctrl-H to pause the messages. If no valid key is pressed within 30 seconds, the default is not to accept.

What it means: after a power event, cable or backplane fault, drive death, or drive shuffle, the controller’s view of the array no longer matches reality. If you know you just added or removed a drive and the message matches exactly, it is expected. If you did not, it is usually a loose connection, a failed backplane slot, or a drive that has stopped responding.

What not to do: do not press Enter to “accept the current configuration” to get the server booting. Adaptec’s own recovery procedure for failed arrays has the administrator reject configuration changes first so the original array parameters can be recorded. Accepting commits the controller’s current, reduced view of the array.

Safe next step: let the 30-second timer expire (or power off). Reseat cables and check drive power with the system off. If every drive comes back and the message disappears, back up immediately. If a drive stays missing, photograph the screen, power down, and stop.

“Booting the Controller Kernel” (system hangs)

Risk: Level 3 if the card never finishes; data usually intact.

The POST sits on this line indefinitely. Adaptec attributes many cases to I/O resource conflicts with onboard motherboard devices, and recommends disabling non-essential onboard devices in the system BIOS to test. On a card that has run for years and suddenly stops here, the more common causes are controller hardware degradation (often heat-related) or a drive that is hanging the firmware during discovery.

Safe next step: power off and disconnect the drives, then see whether the card completes POST with nothing attached. If it does, a drive or backplane is the likely trigger — and the arrays on those drives should be treated as unrecoverable-by-controller until proven otherwise. Do not flash firmware onto a card that is hanging with array members connected.

“Error: Controller Kernel Stopped Running”

Risk: Level 3. Microsemi’s release notes for Series 7 firmware build 32114 list a fix for this error appearing during POST on some systems. On older firmware it can be a firmware defect; on current firmware it points to controller hardware or a device the controller cannot initialize. The arrays are not affected by the message itself — but the next steps people take (firmware flashes, HDA-mode resets, swapping in a different card) can be.

Boot halts because drives are missing

Risk: Level 3. Series 7 cards support a BIOS parameter (set with ARCCONF SETBIOSPARAMS ... BIOSHALTONMISSINGDRIVECOUNT) that stops POST when a specified number of drives is missing. If a server that booted yesterday now halts at the Adaptec banner after listing missing drives, the halt is doing its job. Do not lower the threshold to get past it.

Controller unresponsive after a failed firmware update

Risk: Level 3. Series 7 cards carry active and passive firmware images designed to survive a power loss during an update, but a corrupted flash can still leave a card unable to POST. Adaptec’s documented recovery for Series 6, 7, and 8 is an HDA mode flash using the jumper header on the card, not the older “Mode 0” reset. Perform it with the array drives disconnected. The drives are not needed to recover the controller, and keeping them off the card protects them from anything the recovered firmware does on first boot.

Messages and Prompts in the Ctrl-A Utility (ARC / UEFI HII)

“Initialize Drives” — “During initialization, all data is deleted from the disk”

Risk: Level 4 (destructive). Adaptec’s user guide puts this caution directly in the array-creation procedure: initializing deletes data from the disk. Initialization is required to prepare new drives, and Adaptec’s knowledge base also uses it to clear the state of a replacement drive or to fix an array stuck “degraded” with a member reported as a missing or inconsistent segment. Initializing the wrong drive — for example, a member that dropped out and came back — can take a degraded array to failed or offline. Adaptec says as much in its own article.

Safe rule: never initialize a drive whose serial number appears in the segment list of any logical device, even if that segment is marked missing, until you have a copy of the data or images of every member.

Ctrl+F “Force online” — warning about inconsistent segments

Risk: Level 4 (destructive). In the BIOS utility’s array management screen, Ctrl+F attempts to bring a failed array back online. The utility warns that there may be inconsistent segments, and may offer an “advanced” force online if the standard one fails. Adaptec’s article on recovering failed arrays is careful to say there is no guarantee, and that the technique applies only when all original drives are intact, still attached, and in their original sequence — and that if drives were previously failed, replaced, or hot-spared, you should contact support for other options first.

Why it is dangerous: force online tells the controller to trust the data on every member as current. If one of those members dropped out of the array hours, days, or months ago, its data is stale. Bringing it online means reads may return old data, and a subsequent verify with fix — which Adaptec recommends after a successful force — can “repair” current parity using stale data. You generally get one attempt before the evidence of which member was current is overwritten.

Recreating the array with “Skip Initialization” / “Quick Init”

Risk: Level 4 (destructive). Adaptec’s user guide states that when more than one drive fails in the same array (outside RAID 6 and 60), the data normally can’t be recovered, but that you may be able to recover it by forcing the logical drive online or by recreating the logical drive without the initialization step. The knowledge base procedure has the administrator first record array type, capacity, stripe size, and members, then initialize the members and recreate the array.

Read that sequence carefully: it initializes the drives — erasing the existing metadata — before the recreation. If any recorded parameter is wrong, if a member was in a different position than the utility showed, if the array used non-default settings, or if the array had been expanded or migrated, the recreated array will be misaligned and the original map is already gone. Adaptec’s own warning notes that if the drives belong to more than one array, the procedure destroys the other arrays too.

Separately, “Quick Init” at array creation skips the build process and is recommended by Adaptec only for new drives. Arrays that were created with Quick Init and never verified can carry parity inconsistencies that surface only during a rebuild.

“Controller Mode” setting

Risk: Level 4 when array members are attached. Changing between RAID, HBA, Auto Volume, and Simple Volume modes changes what the controller does with every drive at the next boot. Changing a card that holds arrays into HBA mode, or connecting existing data drives to a card that is in Auto Volume or Simple Volume mode, are the two most common ways this setting damages data. See controller modes above.

“Uninitialize Drives”

Risk: Level 4 (destructive). Removes Adaptec metadata from the selected drives. In the default Expose RAW mode, the drives are then presented to the operating system as raw disks — where the next click in Disk Management or the next Linux installer may write to them.

Clearing preserved controller cache

Risk: Level 4 (destructive). When an array goes offline while writes are sitting in the controller’s cache, Series 7 firmware can preserve that cache instead of discarding it. The Ctrl-A utility and arcconf both offer a way to clear preserved cache. Clearing it throws away the writes the controller was protecting — which may include file system metadata. If a preserved-cache condition exists, the right move is almost always to restore the missing array members first so the cache can be flushed to the drives it belongs to.

What arcconf Output Is Telling You

arcconf is the command-line tool for Series 7 cards. ARCCONF GETCONFIG 1 LD (logical devices), GETCONFIG 1 PD (physical devices), GETCONFIG 1 AL (everything), GETSTATUS 1 (running tasks), and GETLOGS are read-only and safe to run on a failing array. Save their full output to a file on a different machine before anyone changes anything. ARCCONF SAVESUPPORTARCHIVE captures a more complete diagnostic bundle.

“Status of logical device” values

Status	Risk	What it means
Optimal	Level 1	All members present and consistent.
Suboptimal, Fault Tolerant	Level 2	A RAID 6 or RAID 60 array has lost one member. Adaptec uses “suboptimal” because the array can still survive one more failure. Back up and replace the drive; this is the calmest failure state you will see.
Degraded	Level 2	A redundant array has lost as many members as it can tolerate (for example, one drive in RAID 5, or a second drive in RAID 6). Any further failure takes the array offline.
Impacted	Level 1–2	The array is usable with reduced performance, usually because a build or verify has not completed. An array that was created with a build/verify and is still running it will show Impacted. An array that shows Impacted long after creation may never have completed its initial build — which matters if it later needs a rebuild.
Failed / Offline	Level 3	More members are missing or failed than the RAID level can tolerate. The controller will not present the volume. The data is still on the drives.

“Failed stripes : Yes”

Risk: Level 2, with confirmed data loss inside the array. This line in the logical device output means the Bad Stripe Table has entries: specific stripes the controller could not reconstruct and is now blocking. The array may be online and “Optimal” at the same time. Files that touch the blocked stripes return read errors. Adaptec states that bad stripes cannot be cleared while the array is kept, and that workarounds tend to either recreate the same entries or expose the operating system to invalid data. If the files in those stripes matter, the array needs to be reconstructed offline from drive images, where each member’s readable data can be evaluated sector by sector. maxView reports the same condition as a bad stripe event.

“Segment N : Missing”

Risk: Level 2 or 3, depending on how many. The segment list under each logical device names every member by position, connector, device, and serial number. Present means the member is in the array; Missing means the controller no longer has it. Record which serial number belongs in each missing position — that information is on the screen now and may not be later. If a drive shows up as Ready in the physical device list while its serial is listed as a missing segment, it has dropped out of the array; do not initialize it.

“Initializing Channel 0, Device 7 will erase its metadata, the section where all the logical device definition data is stored”

Risk: Level 4 (destructive). This is arcconf’s confirmation prompt for ARCCONF TASK START <Controller#> DEVICE <Channel# ID#> INITIALIZE, followed by Are you sure you want to continue? On a new replacement drive, answering yes is routine. On any drive that was ever part of the failed array, it removes that member’s copy of the array map.

arcconf commands that write to your drives

Command	What it does	Risk on a failed array
SETSTATE <c> LOGICALDRIVE <n> OPTIMAL [ADVANCED]	Forces a failed logical drive online	Level 4 — trusts every member, including stale ones
SETSTATE <c> DEVICE <ch> <id> HSP LOGICALDRIVE <n>	Assigns a hot spare, which starts a rebuild	Level 4 if surviving members have media errors
SETSTATE <c> DEVICE <ch> <id> DDD	Forces a drive to failed	Level 4 if it is the wrong drive
TASK START <c> DEVICE <ch> <id> INITIALIZE	Erases the drive’s Adaptec metadata	Level 4
TASK START <c> LOGICALDRIVE <n> VERIFY_FIX	Verifies and rewrites parity	Level 4 after a force online — can overwrite current parity with stale data
UNINIT	Removes Adaptec metadata	Level 4
DELETE <c> LOGICALDRIVE <n>	Deletes a logical drive	Level 4
SETCONFIG <c> DEFAULT	Deletes logical drives, resets disks to Ready and settings to defaults	Level 4
SETCONTROLLERMODE	Changes RAID / HBA / Auto Volume / Simple Volume mode	Level 4 with members attached
PRESERVECACHE <c> CLEAR ...	Discards preserved cache	Level 4
CREATE <c> LOGICALDRIVE ...	Creates a new logical drive	Level 4 on existing members

maxView Storage Manager Messages

maxView Storage Manager is the browser-based management tool that shipped with Series 7, replacing the older Java-based Adaptec Storage Manager. Its event log and email notifications are where most administrators first learn that something is wrong.

“Logical device is degraded: controller 1 (…), logical device 0 (…) … Repair as soon as possible to avoid data loss.”

Risk: Level 2. The standard maxView warning when a redundant array loses a member. The message includes the controller model and slot, the logical device number and name, and its unique ID. On Adaptec Storage Manager systems the equivalent entries read Logical device is degraded followed by Failed drive: controller 1, channel 0, …, and when the array recovers, Logical device is normal. A stale degraded warning can linger in maxView after a successful rebuild; restarting the maxView service clears the display. A degraded warning you did not expect, however, deserves a backup before anything else.

Bad stripe reported against a logical device

Risk: Level 2 with confirmed unreadable data. maxView reports the same Bad Stripe Table condition that arcconf shows as Failed stripes : Yes. On Adaptec’s OEM-branded management software, the event text reads One or more logical devices contain a bad stripe. See Failed stripes above.

Logical device failed / physical drives all show Optimal

Risk: Level 3. A recurring forum question: every physical drive shows Optimal, yet the logical device shows Failed. That combination typically means the drives are healthy but the array’s membership record no longer matches — a member dropped out and rejoined, drives were reconnected in different positions, or an expander did not present all drives in time. Adaptec has a specific article for the last case: some expanders do not finish scanning drives before the controller firmware looks for them, and the logical device can be marked failed as a result; the fix is an expander minimum-scan-duration setting. That is a configuration problem, not drive failure — which is exactly why it should not be “fixed” with a force online or a recreate.

maxView “Force Online” button

Risk: Level 4. Available in maxView version 1.04 build 20853 and later under the Logical Device group on the ribbon. Same operation and same risks as Ctrl+F in the BIOS utility.

“Set drive state to failed” before removing a drive

Risk: Level 1 when used as intended. Adaptec recommends marking a drive failed in software before pulling it, so the controller and management tools stay in sync. Quick pull-and-replace without this step is one documented cause of an array stuck “degraded” with every member reporting optimal.

Linux Kernel Messages (aacraid Driver)

Series 7 cards are driven by the in-kernel aacraid driver, which identifies them by PCI ID 9005:028c and labels them internally as “Adaptec PMC Series 7 (Denali),” sharing an initialization path with Series 8. The strings below are taken from the current mainline kernel source; exact wording and the device prefix vary with kernel version.

“aacraid: Host adapter abort request” / “Outstanding commands on (…)”

Risk: Level 2, escalating. The SCSI layer asked the driver to abort a command that did not complete in time. An occasional abort can be a single slow drive. Repeated aborts, especially during rebuilds or heavy I/O, usually mean a drive is failing slowly (long internal retries), a cable or backplane is marginal, or the controller itself is overheating or unstable. Owners of secondhand 71605 cards running non-standard firmware have reported these aborts and resolved them by moving to the last public firmware build.

“aacraid: Host bus reset request. SCSI hang ?” and “Adapter health – N”

Risk: Level 3 if it repeats. The error handler escalated past aborts to a bus reset, then checked the controller’s health. A non-zero health value means the firmware reported a problem. If the reset succeeds, I/O resumes; if it does not, the kernel may offline the devices behind the controller.

“adapter kernel panic’d N.” / “aac_fib_send: adapter blinkLED 0x…” — “Usually a result of a serious unrecoverable hardware problem”

Risk: Level 3. The controller’s own firmware crashed. The driver will attempt resets, logging lines such as Controller reset type is, Issuing IOP reset, IOP reset failed, Issuing SOFT reset, or ARC Reset attempt failed. At startup, the related failures are adapter self-test failed, adapter monitor panic, and adapter kernel failed to start. The driver’s own comment describes the blink-LED condition as usually a serious unrecoverable hardware problem. At this point, the controller is the failed component — and the drives should be removed and protected, not moved to “just any” replacement card.

“Device offlined – array failed” / “Device offlined – array deleted”

Risk: Level 3 (“array failed”) or Level 4 aftermath (“array deleted”). These are printed by the aacraid driver when the controller notifies it that a container (array) has changed state. Array failed means the logical device went offline — subsequent reads produce generic SCSI messages such as rejecting I/O to offline device, and filesystems on the array will error out or remount read-only. Array deleted means the logical device was removed from the controller configuration. If nobody on your team deleted an array, find out who or what did before anything else changes.

“Backup Unit had cache data loss” / “DDR cache data recovered successfully”

Risk: Level 2–3 (“data loss”) or Level 1 (“recovered successfully”). These messages report what the flash backup module did after a power loss. DDR cache data recovered successfully means the AFM-700 preserved the cache contents through the outage and restored them. Backup Unit had cache data loss means writes that the operating system believed were complete were not saved. File system damage on the next mount is possible even though every drive is healthy. Do not run repair tools against the volume until you understand what was lost; a repair can make a small inconsistency much larger.

“aac_fib_send: first asynchronous command timed out”

Risk: Level 1–2. The driver’s own text attributes this to a PCI interrupt routing problem and suggests a motherboard BIOS update or safe-mode kernel options. It is a platform compatibility problem rather than a data problem — but it can make a healthy array appear to vanish during an OS upgrade.

Windows Event Log Messages (arcsas Driver)

On Windows, Series 7 cards use the Adaptec arcsas storport miniport driver and appear as \Device\RaidPortN. Windows logs storage timeouts and resets under the driver’s name as the event source.

Event	Typical text	Risk	What it usually means
Event ID 129, source arcsas	Reset to device, \Device\RaidPort0, was issued.	Level 2	A command timed out and Windows reset the device. Isolated: a slow drive or transient. Repeating: a failing member, cabling, or an overheating controller.
Event ID 11, source arcsas	The driver detected a controller error on \Device\RaidPort0.	Level 2–3	A controller-reported error. Check the maxView event log for the underlying cause at the same timestamp.
Event ID 9, source arcsas	The device, \Device\RaidPort0, did not respond within the timeout period.	Level 2–3	The controller stopped responding within Windows’ timeout. Frequent entries often precede a controller hang.
Event ID 143, source arcsas	The device \Device\RaidPort0 is attempting to use more than 8 buses, which exceeds the supported maximum.	Level 1	Documented by Adaptec for Series 7 and Series 8. A driver/OS reporting quirk, not a data problem.
Disk Management	“You must initialize a disk before Logical Disk Manager can access it.”	Level 4	An unknown disk has appeared. After an array failure on a Series 7 card in Expose RAW mode, this may be a former array member. Cancel.

Cache Protection (AFM-700) Messages and Failure States

“Write-through (temporary disabled)”

Risk: Level 2 (performance) — with a firmware twist. Normally this status appears while the supercapacitor is charging or when the backup unit is not healthy, and write-back caching is suspended to protect data. On Series 7 firmware build 32106 (April/May 2016), integrators reported that the status could become permanent regardless of the hardware’s condition, with the reported workaround being a return to build 32084 or earlier. Build 32114’s release notes separately list a fix for write cache that could not be enabled when no flash module was attached. If you see this status with a healthy AFM-700, check the firmware build before replacing hardware.

Supercapacitor over-temperature

Risk: Level 2. Microsemi’s release notes state that when the supercapacitor exceeds its temperature threshold, the backup unit is disabled, its charger turns off, and write caching is disabled for logical drives until the temperature returns to normal. Repeated occurrences are a strong sign the card is not getting the airflow it needs — which also stresses the controller itself.

Supercapacitor removed or inserted while charged

Risk: Level 2–3. Adaptec warns never to remove or insert a charged supercapacitor module, and documents that replacing it while the system is running can put the flash module into a FATAL state. Hot-plugging can also produce odd status transitions (Adaptec’s notes give “Preparing to Dead to Ready” as an example). If this happened immediately after a power loss, the preserved cache may be at risk.

Series 7 Firmware History and Known Issues

All Series 7 cards share a firmware line (version 7.5.0). Knowing which build a card runs helps explain behavior that looks like hardware failure. ARCCONF GETVERSION reports it.

Build	Date	Notes
30855+	2013–2014	Adds the “RAID: Expose RAW” and “RAID: Hide RAW” controller modes.
32033	August 2014	Adds Simple Volume controller mode. At least one storage integrator advised customers to move past this build, citing multiple drives reported degraded at once when they were not defective and volumes disappearing from the operating system.
32084	2015	Referenced by integrators as the last build before the write-through issue below.
32106	April–May 2016	“2016.rel1” Series 7 maintenance release. Reported by integrators to leave write cache stuck in “Write-through (temporary disabled)” on some 71605 cards.
32114	February–March 2017	Patch release. Release notes list fixes for RAID 5 and RAID 6 logical drives that would not rebuild; a firmware stack overflow caused by excessive error handling from a faulty drive; write cache that could not be enabled without a flash module; a parity-cache-log issue that could cause a controller assert; and “Error: Controller Kernel Stopped Running” during POST.
32118	2018	The last Series 7 build published on Adaptec’s public support site.
7.6-0 build 32136	OEM	Found on many secondhand 71605 cards; not a public Adaptec release. Users have reported aacraid aborts on this build and rolled back to 32118.

If a failing Series 7 card is on build 32114 or earlier and a rebuild is refusing to complete, the rebuild fix in 32114 is relevant. Updating firmware while an array is degraded or failed, however, adds risk: a failed update or a changed behavior on first boot happens with your only copy of the data attached. The safe order is data first, firmware second.

The Series 7 Failure Patterns That Bring Arrays to a Lab

The second drive that nobody knew about

A RAID 5 loses a drive, the alarm is silenced or email alerts were never configured, and the array runs degraded for weeks or months. When a second drive fails, the controller marks the array failed. The first drive’s data is stale; the second drive’s data is current but partially unreadable. Force online brings the stale drive back; a rebuild tries to read the unreadable one. Either can finish the array off. Offline reconstruction uses the current drive’s readable sectors and parity from the others, and uses the stale drive only where it is provably still accurate.

The controller that cooked

Series 7 cards run hot and assume server-grade airflow. In tower cases, NAS builds, and closets with failed fans, they overheat. Symptoms range from aborts and resets in the OS logs to arrays dropping members and, eventually, a card that no longer finishes POST. The drives are usually fine. The risk is what happens next: a replacement card from the wrong generation, a firmware flash on a dying card, or a mode change on a used card.

The upgrade to Series 8

A Series 7 card dies or is retired, and a Series 8 card (such as the 81605ZQ) goes in because it is “the next Adaptec.” Every drive shows up as Ready, no arrays exist, and there is no import option. One administrator documented exactly this migration — 11 drives, a 71605 on build 32118, an 81605ZQ, both in Expose RAW mode — with no foreign-configuration option anywhere. The data is intact. What destroys it is the next step: initializing, creating, or letting Windows initialize the raw disks.

The used card in the wrong mode

Secondhand 71605s are popular in home labs and budget servers, often bought as “HBA mode” cards. Existing data drives are connected, and the card — actually in Simple Volume or Auto Volume mode — treats them according to that mode on first boot. Forum reports describe drives that were unmountable afterward, followed by file system repair attempts that overwrote primary superblocks.

The array that failed because the backplane was slow

Every drive is healthy, but the logical device is marked failed after a reboot because an expander backplane did not present all drives before the controller looked for them. Adaptec documents a minimum scan duration setting for this. It is fixable without touching data — unless someone “fixes” it first with a force online or a recreate.

The bad stripe nobody noticed

A rebuild completes, the array returns to Optimal, and Failed stripes : Yes sits quietly in the arcconf output. Months later, a database or VM disk hits the blocked region. The array is healthy by every visible measure, but specific data is gone from the controller’s point of view — and the only path to it is reconstruction from images, where each member’s sectors can be evaluated independently.

What Adaptec’s Documentation Says — and Where We Differ

Adaptec’s manuals and knowledge base are accurate about what the controller can do. They are written for a world where the administrator has current backups and the goal is getting the server back in service. Our starting assumption is different: the backups are missing, incomplete, or older than the data that matters.

• RAID 0 failure: the user guide says that when a drive fails in a RAID 0 array, the data can’t be recovered, and to restore from backup. That is true of the controller. It is not true of the drives — a failed RAID 0 member can often be repaired or imaged in a cleanroom, and the stripe reassembled offline.
• Multiple failures in one array: the guide says the data normally can’t be recovered, but may be by forcing the array online or recreating it without initialization. Both techniques exist; both are single attempts that write to every member. We image first, then reconstruct from copies, so a wrong guess costs nothing.
• Force online: Adaptec itself limits it to arrays whose original drives are all intact, attached, and in original order, and recommends calling support if drives were previously failed or replaced. In practice, that exception covers most of the failed arrays we see.
• Bad stripes: Adaptec says there is no way to clear them while keeping the array and that data in the stripe is incomplete or inconsistent. Offline, the stripe is not a single unit: each member’s copy of that region can be read, compared, and in many cases partially or fully recovered.

What to Do Right Now

1. Stop writing. If the volume is still online and degraded, copy the most important data off it before any rebuild. If it is offline, shut the server down cleanly.
2. Capture the evidence. Photograph every POST and BIOS utility screen. If the OS is up, save ARCCONF GETCONFIG 1 AL, ARCCONF GETLOGS, and a support archive to another machine, and export the maxView event log.
3. Record the physical layout. Label every drive with its bay, the controller connector (CN0–CN3 on a 71605), and its position in any expander backplane. Note serial numbers against the segment list.
4. Do not change controllers, modes, or firmware. Do not move drives to another card or generation.
5. Do not connect member drives to a Windows PC or a Linux installer. Raw array members look like blank or corrupt disks to an operating system.
6. Get an evaluation. Ship the drives together. We do not need the server, and in most cases we do not need the controller.

How Gillware Recovers Adaptec Series 7 Arrays

We never operate a failed Series 7 array during recovery. Every member drive is logged by bay, connector, serial number, and firmware, then imaged on write-blocked hardware in our ISO 5 Class 100 cleanroom. Drives with physical damage — heads, motors, PCBs, firmware — are repaired with donor parts before imaging. The originals are shelved; all work happens on the images.

Our in-house reconstruction software, HOMBRE, reads the Adaptec metadata on every image, determines member order, stripe size, RAID level, and parity layout, and identifies which member dropped out first and when. Where metadata has been destroyed by an initialize, uninitialize, or a Windows disk initialization, the array layout is derived from the file system structures across the images instead. The array is assembled virtually, independent of any controller, and the file systems on it — NTFS, ReFS, VMFS, ext4, XFS — are recovered from the assembled volume. More detail on our Adaptec process is on our Adaptec RAID controller recovery page and our RAID data recovery hub.

Every case starts with a free technical consultation with one of our enterprise storage engineers, and you approve all recovery fees in advance.

Series 7 Model Pages

• Adaptec ASR-71605 / 71605Q / 71605E data recovery — heat, 16-port expander configurations, controller-mode traps on secondhand cards, and the 71605-to-81605ZQ upgrade problem.
• Adaptec ASR-7805 / 7805Q data recovery — eight-port failures, error messages, and maxCache SSD-cache risks.
• Adaptec ASR-72405 data recovery — 24-port array failures, error messages, and Windows sleep-state hangs.

Adaptec Series 7 Manuals and Documentation

Adaptec’s documentation for Series 7 was published under the Adaptec, PMC-Sierra, and Microsemi names, and is now maintained by Microchip. Official download locations have moved several times; the copies below are the versions we used for this page.

Document	What it covers
Serial Attached SCSI RAID Controllers Installation and User’s Guide (CDP-00310-01-A, May 2015)	Hardware specifications for every Series 7 model, AFM-700 installation, the Ctrl-A utility, and the “Solving Problems” chapter on drive failures
Adaptec RAID Controller Command Line Utility User’s Guide (arcconf)	Every arcconf command, including setstate, task, uninit, setconfig, setcontrollermode, and preservecache
Series 7 / Series 8 Readme (July 2014)	Controller modes, Simple Volume and Auto Volume limitations, dual firmware images
Series 7 / Series 8 Readme (2016, firmware build 32106)	2016 maintenance release notes and known limitations
Series 6–8 Readme (October 2016)	AFM-700 status LEDs, supercapacitor over-temperature behavior, cross-series AFM migration
Microsemi Adaptec Series 7 Product Brief	Family overview and model lineup
Linux aacraid driver source	The authoritative source for every aacraid kernel message

Adaptec knowledge base articles referenced on this page: configuration change or missing drives during POST; can a failed array be recovered; RAID, Auto Volume, HBA, and Simple Volume modes; HDA mode flash for Series 6, 7, and 8; what “suboptimal” means; array degraded with all members optimal; Windows Event ID 143 (arcsas).

Frequently Asked Questions

Can a Series 8 or SmartRAID controller read my Series 7 array?

No. Series 8 and SmartRAID controllers list Series 7 drives as Ready, not Foreign, and offer no import. Do not create or initialize anything on the new controller. The arrays are still on the drives and can be reconstructed offline.

Can I replace a dead Series 7 card with another Series 7 card?

Usually, with compatible firmware — the configuration is stored on the drives. Before connecting the drives, confirm the replacement card is in RAID mode (not HBA, Auto Volume, or Simple Volume), and do not accept configuration changes at POST until you have confirmed every member is detected. If the old card died during a rebuild or after drives had already dropped out, get the drives imaged first.

My array says “Impacted.” Is my data in danger?

Impacted usually means a build or verify is incomplete and performance is reduced. On its own it is not a failure. If the array has been Impacted since it was created, it may never have completed its initial build, which increases the risk of problems during a future rebuild.

Is it safe to force my failed array online?

Only if every original member is present, healthy, in its original position, and none of them dropped out earlier than the others. That describes a minority of failed arrays. If you are not certain, image the drives first. Force online is a single attempt that can permanently mix stale and current data.

What does “Failed stripes : Yes” mean if my array shows Optimal?

The controller has permanently blocked specific stripes it could not reconstruct. The array works, but files touching those stripes cannot be read through the controller. Recovering them requires offline reconstruction from drive images.

One drive failed in my RAID 0. Is everything gone?

From the controller’s point of view, yes. From a lab’s point of view, often not — the failed drive can frequently be repaired enough to image, and the stripe reassembled from the images.

Should I update the firmware on a failing Series 7 card?

Not while the only copy of the data is attached. Firmware build 32114 fixed real rebuild problems, but a flash on an unstable card, or changed behavior on first boot, happens with your array connected. Recover or back up the data, then update.

Start Your Adaptec Series 7 Recovery

If your Series 7 array is offline, degraded with failing survivors, or showing messages you now recognize from the Level 3 and Level 4 lists above, power the server down and label the drives by bay and connector. Do not initialize, force, recreate, or migrate anything. Our engineers image every member and reconstruct the array without the controller.