Skip to content
View in the app

A better way to browse. Learn more.

Unraid

A full-screen app on your home screen with push notifications, badges and more.

To install this app on iOS and iPadOS
  1. Tap the Share icon in Safari
  2. Scroll the menu and tap Add to Home Screen.
  3. Tap Add in the top-right corner.
To install this app on Android
  1. Tap the 3-dot menu (⋮) in the top-right corner of the browser.
  2. Tap Add to Home screen or Install app.
  3. Confirm by tapping Install.

Recurring Task Aborts / I/O errors / Device Resets on LSI 9305-24i during SMART polls while drives are spinning up (Unraid 7.3.1)

Featured Replies

Hello, I'm new here. Up until now, I've always managed to resolve my "problems" and questions on my own somehow. With this particular issue, however, I'm definitely reaching my limits, as it appears to stem from the OS or kernel itself—"normal" behavior.

Regarding the problem:
Symptoms:

  • Frequent kernel messages of the type:

    • attempting task abort! ... outstanding for ~7000 ms & timeout 7000 ms

    • I/O error, dev sdX, sector 0 op 0x0:(READ)

    • Power-on or device reset occurred

  • These errors often coincide with emhttpd: read SMART /dev/sdX

  • In some cases this also produces real array read errors (md: diskX read error)

  • As far as I can tell, it affects only mechanical storage drives.

Important observations:

  • SMART attributes of all affected drives are completely clean (0 reallocated, 0 pending, 0 UNC, 0 CRC errors).

  • The errors only occur when a drive is in the process of spinning up and a SMART query (or the preceding “is the drive ready?” check) is issued at the same time.

  • When a drive is fully spun down, SMART queries are correctly discarded.

  • When a drive is already fully spun up, SMART queries succeed without problems.

Hardware (already extensively tested):

  • Chassis: Inter-Tech 4424-4U

  • Backplane: Inter-Tech SU-2 (also tested with original backplanes)

  • HBA: LSI SAS9305-24i (Firmware 16.00.12.00)

  • Cables: tested multiple generations (including Delock 85081 and current Delock 86092)

  • PSU: Seasonic Prime 1600W Noctua Edition (previously be quiet! Dark Power Pro 1000W) – power distribution optimized, each backplane on its own peripheral cable

  • Motherboard: ASUS ProArt B550-Creator

  • CPU: AMD Ryzen 9 5950X

  • GPU: nvidia RTX 4000 Ada full size

  • RAM: 128 GB

  • Drives: mainly Toshiba MG10AFA/MG10ACA + older WD Red (WD60EFRX)

Hardware changes (cables, backplane, PSU) did not resolve the issue.

Settings:

  • Default spin-down delay: 3 hours

  • No spin-up groups (intentional – only the needed drive should spin up for Plex etc.)

  • poll_attributes=604800

Software (relevant (?)):

  • cooler control

  • Plex

  • disk location

  • Parity Check Tuning

  • Unassigned Devices and Plus

  • User Scripts

  • Dynamix Cache Directories

Of course, more Docker containers and VMs are supposed to follow, but right now I have little confidence in my system due to the issue at hand.

The problem appears to be a timing issue between the short timeout used by emhttpd/kernel for the SMART readiness check (~7 seconds) and the time some drives + the LSI controller need during spin-up.

Diagnostics are attached.

I’ve also opened a ticket; I’m just hoping for quick, effective assistance.

Many thanks for your help, and best regards.

jarvis-diagnostics-20261013-0746.zip

Solved by Knolte

Your analysis looks correct to me, and there's no sign of a hardware problem.

Here is what happens each time, from the logs:

1. A disk in standby gets a read, and starts to spin up.

2. While it is spinning up, Unraid sends it a power-state/SMART status check. In your logs that command has a 7-second timeout (10 seconds for a few SMART log reads).

3. Your drives take about 9–9.5 seconds to spin up, according to their own SMART Spin_Up_Time. The command times out before the drive is ready.

4. The kernel aborts the command, and the HBA resets the drive. That is the attempting task abort and Power-on or device reset occurred you see.

5. The reset also cancels the normal reads that were waiting for the drive. Those show up as I/O error and, for array disks, md: diskX read error.

Every abort in your logs is one of these status commands. No normal read or write ever timed out, and all affected drives have clean SMART. That is why changing cables, the backplane and the PSU made no difference: the hardware is doing what it is told.

When an array disk returns a read error, Unraid rebuilds the data from parity and writes it back to the disk. No disk was disabled, and there are no write errors, so nothing was lost. You can reset the error counters on the Main page.

To confirm the cause, please disable spin-down for a few days. Then check whether the errors are gone. If you no longer see any task aborts, resets or read errors with spin-down disabled, please create a bug report at https://product.unraid.net. Include your diagnostics and a link to this thread, so that LT can look at the timeout used for this status check.

  • Author

Hello, thank you very much for your feedback. It’s reassuring to know that my analysis was at least somewhat on the right track. :-D I’ll follow your suggestion: I’ll let the HDDs spin for 2–3 days and then send the error report, etc., as you instructed. What still surprises me, though, is that I seem to be the only one (?) with this problem. I can’t find anything about it in the forums. And if the issue originates in the "kernel" itself, why is a 7-second threshold defined there? I’m a total noob when it comes to this sort of thing, so please don't be too hard on me! ;-) One more question. You wrote: "You can reset the error counters on the 'Main' page." -> Where? I couldn't find a counter reset option there. I ended up restarting Unraid.

10 minutes ago, Knolte said:

I can’t find anything about it in the forums.

It could be an issue with the hardware you are, or a combination of the hardware used.

11 minutes ago, Knolte said:

You wrote: "You can reset the error counters on the 'Main' page." -> Where? I couldn't find a counter reset option there. I ended up restarting Unraid.

Yes, I think that no longer exists. Starting the array is the correct way.

  • Author

Hello,

it’s time for an update:

Since I’ve kept the HDDs spinning, no more read errors have occurred. The diagnostic data is attached.

The very helpful support team reached out to me today and pointed out the following:

"We noticed something unusual in your diagnostic data: your 'go' file loads the 'drivetemp' driver (modprobe drivetemp). Every time this driver queries the hard drive temperatures, it sends a SMART command to each drive—with a 10-second timeout. Some of the interruptions in your logs are caused by this command being sent precisely when a drive was in the middle of spinning up. Specifically, these are the interruptions with a 10,000 ms timeout. This driver is not loaded in the default Unraid configuration; we therefore recommend that you remove it permanently. It interferes with how Unraid manages drive spin-up/spin-down and temperature monitoring. Furthermore, a temperature query via this driver can cause a hard drive to wake up from standby mode. Please remove the relevant line from the 'go' file (/boot/config/go) and restart the system."

Many thanks for this tip. I don’t recall exactly why I wanted to read the drive temperatures this way. Could this be related to the 'Cooler Control' Docker container?

Recommended next steps from support:

1. Remove the 'drivetemp' line from the 'go' file, as described above.

2. Re-enable spin-down so that the hard drives can enter standby mode as intended and spin back up when needed.

3. Start the system in "Safe Mode." In Safe Mode, no plugins are loaded. You can select this mode via the boot menu or by navigating to "Main" > "Flash" > "Syslinux Configuration."

4. Use the server as you normally would for one or two days, allowing some hard drives to spin up from standby mode.

5. Send us new diagnostic data and let us know whether the interruptions—characterized by a 7,000 ms timeout—continue to occur.

"If the errors no longer occur in Safe Mode, we can re-enable the plugins one by one to identify which one is causing the issues. If they persist even in Safe Mode, the diagnostic data will help us further narrow down the root cause."

What I have done:

1. I deleted the "drivetemp" line from the "go" file.

2. I re-enabled spin-down and set it to 3 hours.

3. I restarted the server and booted into Safe Mode. (I am using a KVM and selected "Safe Mode" during startup—I assume this is the correct procedure.)

I will leave the server running like this over the weekend and will share the results—along with the diagnostic data—here in the forum on Monday.

In the meantime, here is a complete list of my installed plugins:

- Disk Location

- Community Applications

- Dynamix Cache Directories

- Dynamix System Temperature

- GPU Statistics - Nvidia Driver

- Parity Check Tuning

- Unassigned Devices

- Unassigned Devices Plus

- User Scripts

Alright, now it’s just a matter of waiting and hoping. I’m curious to see if one of these plugins is the culprit.

Many thanks to Lime Technology for their friendly and very fast support—I feel like I’m in good hands.

jarvis-diagnostics-20261016-1517.zip

Edited by Knolte

Thanks for the update, I missed the custom go file, so yeah, removing that and also booting in safe mode is a good retest.

  • Author

Now I know why I enabled drivetemp via a Go file. Cooler Control can no longer read HDD and SSD temperatures! AND the Graphics card ist offline too - Plex not runnig. I will test these settings until around 11:00 a.m. tomorrow. By that time, errors have typically always occurred.

cooler control.PNG

Edited by Knolte

  • Author

Hello, a new update:

After disabling all plugins (safe mode), no further read errors have occurred. While I haven't tested it for very long, I trust Lime Tech when they say that the 7ms and 10ms timeouts are not originating from Unraid itself.

This leaves the plugins as the primary suspects. It would be very helpful if one could simply disable plugins rather than having to uninstall them—that’s a shame. The fact is, one of the plugins is causing a 7ms timeout during a SMART query.

My top suspect: Disk Location (not a Lime Tech plugin, and it performs SMART queries).

The other plugins are either from Lime Tech or, in my view, unrelated to the storage drives.

I have uninstalled Disk Location and will test the system for a few days, then report back here.

I am keeping drivetemp disabled (commented out) in the go file; Lime Tech says that this driver is not recommended, as I'm solving with cooler control - let's see. I think that Docker container is absolutely brilliant, especially since its major update.

Lime Tech suggested (via email) that I use the "Dynamix Auto Fan Control" plugin. Unfortunately, I cannot find (App Store) a plugin by that name. Could someone here perhaps clarify which plugin they are referring to? Alternatively, does anyone know how I can enable Cooler Control to read drive temperatures without using drivetemp (modprobe)?

Many thanks and best regards.

jarvis-safemode-diagnostics-20261017-0830.zip

Edited by Knolte

Tryu FanCtrl Plus, it's more recent, search for that on the app store.

You can temporarily disable a plugin by renaming its .plg file in /boot/config/plugins to, for example, *.back and then rebooting.

  • Author

Hello, after about 4 days—following the deactivation/uninstallation of "drivetemp" (10 s trigger) and "Disk Location" (7 s trigger?)—I can report that I am no longer seeing any read errors on my disks. Furthermore, I’ve noticed—though perhaps it’s just my imagination—that Unraid feels smoother overall.

I would like to express my sincere thanks for the support I received here on the forum (it was fast, well-informed, and effective). "FanCtrl Plus" is exactly what I was looking for; it’s a perfect replacement for "Cooler Control." I also want to extend a big thank you to the official support team—they were equally fast, knowledgeable, and helpful. I’m very glad I chose Unraid.

I’ve also learned a valuable lesson here: don’t install too many things at once! It’s better to add them one by one, let them run for a few days, and see what happens.

Edited by Knolte

  • Author
  • Solution

Hello, here is my final update on this topic. Unraid support summarized it very well, which might be helpful to someone else facing similar issues. As far as I’m concerned, the problem is resolved.

To sum up what we found:

1. The drivetemp driver, loaded from your go file, sent the SMART commands with the 10-second timeout. When one of those reached a disk that was still spinning up, the command timed out and the disk was reset.

2. The commands with the 7-second timeout were not from Unraid. They stopped in Safe Mode and stayed gone after you removed Disk Location, so that plugin is the most likely source.

In both cases the reset also cancelled the normal reads waiting for the disk. That is where the read errors came from. Your disks, cables, backplane and PSU were fine, and no data was lost.

Your lesson is a good one. If the errors ever come back after you add something new, please download the diagnostics before you reboot and send them to us.

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

Account

Navigation

Search

Search

Configure browser push notifications

Chrome (Android)
  1. Tap the lock icon next to the address bar.
  2. Tap Permissions → Notifications.
  3. Adjust your preference.
Chrome (Desktop)
  1. Click the padlock icon in the address bar.
  2. Select Site settings.
  3. Find Notifications and adjust your preference.