September 30Sep 30 Hello everyone,For quite some time I have encountered random shutdowns, unclean Shutdowns and crashes. Although I never put 1 and 1 together, in the early hours this morning it happened again and I do not have a solution to it or a cause for why it is happening. Attached is the text I was posting in a different thread but after feedback a new thread seemed to be the way to go. Diagnostics are attached.This morning I was wondering why my Server wasn’t showing up within my finder on mac, since is happens from time to time that finder throws it out, I wasn’t really concerned with it - After 3 Failed reconnects I was and If it wasn’t for them, after Plex hasn‘t shown any Libraries after starting it - I was.Quick trip to the GUI and I knew what happened: System Crashed, out of the Blue at roughly 12pm last night, right around midnight. I don’t know what happened or why it happened in the first place.I’ve attached the diagnostics but I’d really appreciate any tips or solutions to or for fixing it. Sick of these unclean Shutdowns for months nowCheers and Thanks in AdvanceAlexserver-diagnostics-20260930-1345.zip Edited September 30Sep 30 by deltaexray
September 30Sep 30 The diagnostics are from after the reboot, so they can't show what happened before the crash. The log starts at 04:18, so the server rebooted on its own sometime after the problem, unless someone restarted it manually. The current boot looks clean: no hardware (MCE/ECC) errors are logged, and the board didn't report any error from the previous boot.To find the cause, we need a log from the next crash:1. Go to Settings → Syslog Server and enable Mirror syslog to flash, then Apply. The log will survive a crash or reboot. After the next event, post new diagnostics together with the saved syslog (/boot/logs/syslog-previous or similar).2. Your board has IPMI, so check the BMC System Event Log for entries between about midnight and 04:20 last night. Use the IPMI plugin's Event log page or the BMC web interface. Look for power loss or PSU events, watchdog resets, memory or ECC errors, or temperature events, and post a screenshot of anything from that window.3. Tell us: - When you find the server after one of these events, is it already rebooted (like today), powered off, or frozen and not responding (monitor/IPMI console)? - Is the server on a UPS? - Did the crashes start after a specific change, such as an Unraid update, a BIOS update, new hardware, or new VMs/containers? Or have they happened for as long as you can remember?With the persistent syslog and the BMC event log, we should be able to tell whether this is hardware, power, or software.
September 30Sep 30 Author 5 hours ago, JorgeB said:The diagnostics are from after the reboot, so they can't show what happened before the crash. The log starts at 04:18, so the server rebooted on its own sometime after the problem, unless someone restarted it manually. The current boot looks clean: no hardware (MCE/ECC) errors are logged, and the board didn't report any error from the previous boot.To find the cause, we need a log from the next crash:1. Go to Settings → Syslog Server and enable Mirror syslog to flash, then Apply. The log will survive a crash or reboot. After the next event, post new diagnostics together with the saved syslog (/boot/logs/syslog-previous or similar).2. Your board has IPMI, so check the BMC System Event Log for entries between about midnight and 04:20 last night. Use the IPMI plugin's Event log page or the BMC web interface. Look for power loss or PSU events, watchdog resets, memory or ECC errors, or temperature events, and post a screenshot of anything from that window.3. Tell us:- When you find the server after one of these events, is it already rebooted (like today), powered off, or frozen and not responding (monitor/IPMI console)?- Is the server on a UPS?- Did the crashes start after a specific change, such as an Unraid update, a BIOS update, new hardware, or new VMs/containers? Or have they happened for as long as you can remember?With the persistent syslog and the BMC event log, we should be able to tell whether this is hardware, power, or software.Hello, Thanks for your input, I just wanna say: I'm a bit under the weather and quite sick for a few days now, It might be the reason I haven't thought of any these tips beforehand - My Apologies. Anyway, back to what you've said. Yeah I noticed that too, which led me to kind of a wild goose chase to be honest since after checking the IPMI, btw thank you for that tip!, I was as clueless as before since it wasn't showing any signs of answers to figure out what has happened to it. But after some time I figured out what must have happened to it: Yesterday and the days before that, it ran the Loudness Data Analysis from Plex which it does from time - So nothing knew there. Well the fans picked up quite the RPM after some time, given I had 27° in the room it wasn't something to be concerned about but it kept running like that beyond 12PM, the Schedule for these Plex Operations is 11AM to 11PM, so it should've been off by that time but it wasn't. I can only imagine that it kept running like I had described before well beyond 3AM and after that, It crashed for some reason - The exact reason? I have no idea. Temperature? Wattage? Power? Idk. 1. Syslog Server has now been turned on, after Parity Check finishes I'll reboot and figure out why it's not giving me the option to select the folder to store these logs. 2. Already answered this kind of, the IPMI is not showing anything useful, just the Turn On time that was around 3AM ish. 3. Today it has been completely rebooted and when navigating to the IP to login into the GUI, I was presented with the Main Screen to select which drives go where - So basically the screen where you input the encryption phase / file. The Drives are always in the right spot, that has never changed but it tells me the system has crashed since there is nothing which would explain that behavior. 4. Server is on a UPS which is sufficient to let it power off gently and without any unclean shutdowns. 5. Couple Years ago, like in the 6.X era these things almost never happened, I'd go as far as to say these only started to happen in the 7.2/3 era. But like this? It's the first time. Like the system got stuck a couple of times over the years and needed a full hard reboot but that was mostly because I was dumb. A complete system crash, like this, never happened. Why do I know this? Because this message: Has never appeared before. This is the first time. Before you ask: Parity Check Tuning is installed but it hasn't been turned on nor is it running scheduled Checks. So yeah that's all. I'll add new information when I have 'em. Cheers and Thank you in advance!Alex
October 1Oct 1 5 hours ago, deltaexray said:Hello, Thanks for your input, I just wanna say: I'm a bit under the weather and quite sick for a few days now, It might be the reason I haven't thought of any these tips beforehand - My Apologies. Anyway, back to what you've said. Yeah I noticed that too, which led me to kind of a wild goose chase to be honest since after checking the IPMI, btw thank you for that tip!, I was as clueless as before since it wasn't showing any signs of answers to figure out what has happened to it. But after some time I figured out what must have happened to it: Yesterday and the days before that, it ran the Loudness Data Analysis from Plex which it does from time - So nothing knew there. Well the fans picked up quite the RPM after some time, given I had 27° in the room it wasn't something to be concerned about but it kept running like that beyond 12PM, the Schedule for these Plex Operations is 11AM to 11PM, so it should've been off by that time but it wasn't. I can only imagine that it kept running like I had described before well beyond 3AM and after that, It crashed for some reason - The exact reason? I have no idea. Temperature? Wattage? Power? Idk. 1. Syslog Server has now been turned on, after Parity Check finishes I'll reboot and figure out why it's not giving me the option to select the folder to store these logs. 2. Already answered this kind of, the IPMI is not showing anything useful, just the Turn On time that was around 3AM ish. 3. Today it has been completely rebooted and when navigating to the IP to login into the GUI, I was presented with the Main Screen to select which drives go where - So basically the screen where you input the encryption phase / file. The Drives are always in the right spot, that has never changed but it tells me the system has crashed since there is nothing which would explain that behavior. 4. Server is on a UPS which is sufficient to let it power off gently and without any unclean shutdowns. 5. Couple Years ago, like in the 6.X era these things almost never happened, I'd go as far as to say these only started to happen in the 7.2/3 era. But like this? It's the first time. Like the system got stuck a couple of times over the years and needed a full hard reboot but that was mostly because I was dumb. A complete system crash, like this, never happened. Why do I know this? Because this message: Has never appeared before. This is the first time. Before you ask: Parity Check Tuning is installed but it hasn't been turned on nor is it running scheduled Checks. So yeah that's all. I'll add new information when I have 'em. Cheers and Thank you in advance!AlexUnraid always performs a parity check after an unclean shutdown (sudden reboot or power loss event)
October 1Oct 1 Author 8 hours ago, MowMdown said:Unraid always performs a parity check after an unclean shutdown (sudden reboot or power loss event)HelloI do know that, that’s not even a bad thing. I just say this notification for the first time that said “System Crash“ directly
October 1Oct 1 1 hour ago, deltaexray said:HelloI do know that, that’s not even a bad thing. I just say this notification for the first time that said “System Crash“ directlyNote that the message is from the Parity Check Tuning plugin - not the core Unraid system. The plugin has detected that no attempt was made to tidily shut down the server so it assumes that it crashed.
October 1Oct 1 Author 15 minutes ago, itimpi said:Note that the message is from the Parity Check Tuning plugin - not the core Unraid system. The plugin has detected that no attempt was made to tidily shut down the server so it assumes that it crashed.That actually makes a lot more sense, Thank you for clearing this up the right way! Also, I think I should setup Nut to have better monitoring and control over the UPS?
October 14Oct 14 Author Hello everyone, Well it happend again, this morning, around the same time as last time, my server just went "quite" and within a few seconds it completely rebooted. Syslog is attached to please feel free to tell me your point of view Thanks syslog-previous-2026-10-14
October 14Oct 14 The log ends at 03:54:35 with nothing unusual: no errors, no crash and no shutdown. The server was idle, with the disks spun down, so this time it wasn't related to load or heat. You also saw it go quiet and restart within seconds. An instant reset with nothing in the log usually points to hardware or power, because a software crash normally logs something first.Please post:1. The diagnostics from now, i.e. from the boot after this reset. If the board or CPU recorded a hardware error, it shows up in the next boot.2. The IPMI/BMC event log entries around 03:50–04:00 this morning, even if they look unimportant.If both show nothing, the PSU is the most likely suspect. A UPS doesn't protect against a PSU that fails or trips internally. If you can, test with a different PSU.Two unrelated things:- On Oct 2, disk3 had a series of read errors. Unraid corrected them from parity, but please run an extended SMART test on that disk and post the SMART report when it's done.- Every line in your syslog appears two or three times. You seem to have the remote syslog server set as well, probably to this server. Clear the Remote syslog server field and keep only Mirror syslog to flash.
October 17Oct 17 Author On 10/14/2026 at 5:01 PM, JorgeB said:The log ends at 03:54:35 with nothing unusual: no errors, no crash and no shutdown. The server was idle, with the disks spun down, so this time it wasn't related to load or heat. You also saw it go quiet and restart within seconds. An instant reset with nothing in the log usually points to hardware or power, because a software crash normally logs something first.Please post:1. The diagnostics from now, i.e. from the boot after this reset. If the board or CPU recorded a hardware error, it shows up in the next boot.2. The IPMI/BMC event log entries around 03:50–04:00 this morning, even if they look unimportant.If both show nothing, the PSU is the most likely suspect. A UPS doesn't protect against a PSU that fails or trips internally. If you can, test with a different PSU.Two unrelated things:- On Oct 2, disk3 had a series of read errors. Unraid corrected them from parity, but please run an extended SMART test on that disk and post the SMART report when it's done.- Every line in your syslog appears two or three times. You seem to have the remote syslog server set as well, probably to this server. Clear the Remote syslog server field and keep only Mirror syslog to flash.Hello @JorgeB and everyone else, Work got me tied up this week so excuse the late reply. I checked your mentioned and have attached the Diagnostics and A Screenshot of the IPMI readout,tho it hasn't been much. server-diagnostics-20261016-1939.zipserver-smart-20261017-1406.zipRegarding the unreleated things: I'm aware of disk3 and it's related issues, a replacement drive has been ordered and is currently being precleared before inserted as a "new" disk3. However thank you for pointing out the fact that unraid corrected them from parity, I'll double check them. Note: SMART short self test completed successful while the SMART extended-self test failed almost instantly. SMART report is attached. The remote syslog server option has been diasabled, thank you for pointing that out too - I wasn't really sure on how to set it up so I just went with everything. Cheers
October 18Oct 18 Replacing disk3 is the right next step. It reports many reallocated and pending sectors, and also several extended tests failed with read errors.Use the normal disk replacement procedure to rebuild onto the replacement in the disk3 slot. For the resets, the new diagnostics show an unclean boot on October 14 at 05:02, but no hardware error or kernel crash record explains it. Please attach a text export of the BMC System Event Log covering that boot, allowing for any difference between its clock and Unraid’s. Keep syslog mirroring enabled. If the BMC provides no explanation, a test with a known-good PSU remains reasonable, but the PSU is not a confirmed cause.
October 18Oct 18 Author 5 hours ago, JorgeB said:Replacing disk3 is the right next step. It reports many reallocated and pending sectors, and also several extended tests failed with read errors.Use the normal disk replacement procedure to rebuild onto the replacement in the disk3 slot.For the resets, the new diagnostics show an unclean boot on October 14 at 05:02, but no hardware error or kernel crash record explains it.Please attach a text export of the BMC System Event Log covering that boot, allowing for any difference between its clock and Unraid’s. Keep syslog mirroring enabled. If the BMC provides no explanation, a test with a known-good PSU remains reasonable, but the PSU is not a confirmed cause.Hey @JorgeB Thank you for your answer, Replacement is currently being precleared and after that, I'll swap the old disk3 out for the new one and let it rebuild through parity. Idk tbh how it is always an unclean shutdown or an unclean boot, everything is setup so it's just "working" but for a little while now, everything is acting so weird. Also: I'm with you on the PSU subject but why is it then "crsahing" twice at the same time roughly two weeks apart? If it were a faulty PSU it surely wouldn't be so precise apart wouldn't it? For better or for worse, I can't find any extendes BMC System Event Log that is covering the boot sequence, IDK abou the Supermicro IPMI but I can't find anything other than what I've already found. On the other hand, I found out my BIOS and BMC Firmware were out of touch, so I'll update them too when rebooting for the HDD swap. Cheers and ThanksAlex
November 3Nov 3 Author Update Time: Hello everyone, it has been a while but I've got some news and updates. First: The Disk3 has been replaced & functions now the way it's been supposed to all along. So that issue has been solved. Now to the actual problem: Idk for sure what the problem has been with the random shutoffs and random restarts but for now I have ordered a new battery for the UPS. I think that is probably the issue since the last time it happened, that was about a week ago, I noticed how the UPS "clicked" i.e the circuit breaker inside of the ups clicked or has worked, in a way, after that the Server started again. So I guess it's either the self-battery check on the UPS failing, hence the new start or something's wrong with the UPS itself. The Battery replacement is cheaper, hence why I am going with that first. When that doesn't help and the issue continues I'll buy a new and more powerful one - I need more capacity anyway. While on the topic of this: Why is it that when I shutdown the Server from the UnRaid Main tab and it has turned off, it displays an unclean Shutdown message after starting it again? It will ALWAYS run a Parity Check after Starting it again tho it has been shutdown the correct way. Am I missing something or doing something wrong? Would gladly take any advice on this. CheersAlex
November 3Nov 3 Good to hear disk3 is sorted.About the resets, the UPS is a good lead, and the timing supports it. The three events were on September 30, October 14 and about a week ago, so roughly 14 days apart and always in the early morning. Many APC units run an automatic self-test every 14 days (336 hours). During the self-test the UPS switches the load to the battery. If the battery is weak and cannot carry the load, the output drops for a moment and the server resets. The board then powers it back on. That fits what you saw: the UPS clicks, the server restarts within seconds, and nothing is logged. It also explains the regular timing you asked about earlier.To confirm it, please post the output of this command from the Unraid terminal:apcaccessThe most useful fields are LASTXFER (the reason for the last transfer to battery), SELFTEST, STESTI (the self-test interval), LOADPCT, BCHARGE, TIMELEFT and BATTDATE. With an EPYC, a 24-port HBA full of disks and a GPU, also check that LOADPCT leaves enough headroom. An old battery near its limit may handle a short power blip but fail under a full load.Until the new battery arrives, you can turn off the UPS automatic self-test, or at least note when the next one is due. After you replace the battery, run a manual self-test while the server is on, at a time when a reset would not cause problems. If the server stays up, the cause is very likely found. Keep syslog mirroring to flash enabled until then. If it happens again, post the diagnostics from the next boot.About the shutdown: when you shut down from the GUI and the next boot still reports an unclean shutdown, usually the array did not stop before the shutdown time-out expired. Unraid then forces the shutdown, and the next boot flags it as unclean and starts a parity check. A VM, a container or an open file that keeps a disk busy can cause this. To check:1. With the server running, stop the array manually from the Main page and note how long it takes. Watch for any VM that is slow to shut down, for example the one using the GPU.2. Go to Settings → Disk Settings and make sure Shutdown time-out is longer than that time, with some margin.3. If it happens again, look in /boot/logs on the flash drive. When the time-out is hit, Unraid saves a diagnostics file there during the shutdown. Post it, because it shows what blocked the array stop.
November 3Nov 3 Author 33 minutes ago, JorgeB said:Good to hear disk3 is sorted.About the resets, the UPS is a good lead, and the timing supports it. The three events were on September 30, October 14 and about a week ago, so roughly 14 days apart and always in the early morning. Many APC units run an automatic self-test every 14 days (336 hours). During the self-test the UPS switches the load to the battery. If the battery is weak and cannot carry the load, the output drops for a moment and the server resets. The board then powers it back on. That fits what you saw: the UPS clicks, the server restarts within seconds, and nothing is logged. It also explains the regular timing you asked about earlier.To confirm it, please post the output of this command from the Unraid terminal:apcaccessThe most useful fields are LASTXFER (the reason for the last transfer to battery), SELFTEST, STESTI (the self-test interval), LOADPCT, BCHARGE, TIMELEFT and BATTDATE. With an EPYC, a 24-port HBA full of disks and a GPU, also check that LOADPCT leaves enough headroom. An old battery near its limit may handle a short power blip but fail under a full load.Until the new battery arrives, you can turn off the UPS automatic self-test, or at least note when the next one is due. After you replace the battery, run a manual self-test while the server is on, at a time when a reset would not cause problems. If the server stays up, the cause is very likely found. Keep syslog mirroring to flash enabled until then. If it happens again, post the diagnostics from the next boot.About the shutdown: when you shut down from the GUI and the next boot still reports an unclean shutdown, usually the array did not stop before the shutdown time-out expired. Unraid then forces the shutdown, and the next boot flags it as unclean and starts a parity check. A VM, a container or an open file that keeps a disk busy can cause this. To check:1. With the server running, stop the array manually from the Main page and note how long it takes. Watch for any VM that is slow to shut down, for example the one using the GPU.2. Go to Settings → Disk Settings and make sure Shutdown time-out is longer than that time, with some margin.3. If it happens again, look in /boot/logs on the flash drive. When the time-out is hit, Unraid saves a diagnostics file there during the shutdown. Post it, because it shows what blocked the array stop.Hello @JorgeB and everyone, Yes, it went smooth as butter when it came to replacing and rebuilding. I do think so too, by now, at least since it is always the same time for it to happen. The Result of the apcaccess command is the following:APC : 001,029,0692DATE : 2026-11-03 13:23:15 +0100HOSTNAME : ServerVERSION : 3.14.14 (31 May 2016) slackwareUPSNAME : ServerCABLE : USB CableDRIVER : USB UPS DriverUPSMODE : Stand AloneSTARTTIME: 2026-10-31 20:35:56 +0100MODEL : Ellipse PROSTATUS : ONLINELOADPCT : 52.0 PercentBCHARGE : 100.0 PercentTIMELEFT : 5.4 MinutesMBATTCHG : 75 PercentMINTIMEL : 75 MinutesMAXTIME : 75 SecondsOUTPUTV : 216.0 VoltsDWAKE : -1 SecondsLOTRANS : 165.0 VoltsHITRANS : 285.0 VoltsALARMDEL : 30 SecondsLINEFREQ : 50.0 HzNUMXFERS : 0TONBATT : 0 SecondsCUMONBATT: 0 SecondsXOFFBATT : N/ASTATFLAG : 0x05000008SERIALNO : G355N17209END APC : 2026-11-03 13:23:18 +0100 I don't think you'll find what you're looking for within this data. Also I cannot turn of selfttests or anything with this UPS, since it's not that smart. Eaton Ellipse PRO 650VA/400W. I'll do my best, battery is supposed to arrive this Thursday and I will make the change either the same day or Friday. I do know I'm flying really close to the sun with the Server Power Draw AND a switch attached to it, which is not a lot but probably 20Watts with the AP included?. So yeah. EDIT: If anyone has a good recommendation for a UPS that quiet and smart, I'll gladly take them. I will do this right about now to figure this out, it was set to 90 seconds by default. I'll test it and report back.Cheers and Thank youAlex Edited November 3Nov 3 by deltaexray
November 3Nov 3 Thanks for the output. One correction to my last post: your UPS is an Eaton, not an APC, so the APC 14-day default doesn't apply directly. The idea is the same, though. Many UPS units run an automatic battery test on their own schedule, and the regular early-morning timing still fits that. With an Eaton on USB, apcupsd doesn't get the self-test or last-transfer data, so those fields are missing from the output. That is expected and doesn't mean anything is wrong. The output does show something important: the UPS is small for this server. LOADPCT is 52% of 400 W, about 210 W, and that reading was taken with the server mostly idle. With the EPYC, all disks spinning, the GPU VM running, and the switch and AP on the same UPS, the load can reach or pass the 400 W rating. TIMELEFT is only 5.4 minutes even on a full charge. If the UPS switches to battery during a test, with an old battery and that much load, the output can drop for a moment and the server resets. That matches the click and the restart you saw.The new battery should help, but I would still plan to replace the UPS. Also check the Ellipse PRO datasheet for its output waveform on battery. If it is a stepped sine wave rather than a pure sine wave, some PSUs with active PFC can also drop out at the moment of transfer.When the new battery is fitted and fully charged, test it at a time when a reset won't cause problems: with the server running, unplug the UPS from the wall. Your UPS settings currently use 75 minutes for the runtime threshold, which is more than the UPS can supply. So as soon as it goes on battery, Unraid starts a clean shutdown. That is fine for a UPS this size, and the test then checks two things at once: that the battery carries the load, and that the shutdown finishes cleanly. If the next boot doesn't report an unclean shutdown, both work. Because of the short runtime, keep the shutdown time-out well below about 4 minutes. 90 seconds is good if the array stops within that time.For a new UPS I can't recommend a specific brand, but these are the points to look for: - Line-interactive with pure sine wave output. - A watt rating, not just VA, where your normal load is about half of it or less. For this server, that means at least about 900 to 1000 W. - A USB connection that works with apcupsd or the NUT plugin, so Unraid can monitor it and shut down cleanly. - If noise matters, check the fan specification. Some units only run the fan while on battery or while charging. Please post the results after the battery swap and the unplug test. If the server resets again in the early morning, post the diagnostics from the next boot together with the mirrored syslog.
November 3Nov 3 Author I have to be the one for Thanking you Yes, even tho mine is an Eaton Unit, I do believe that too. Hence the new battery. I'll keep that in mind, a longer shutdown time for the array while the battery is draining would end up in the same situation, I realized that too a couple of minutes ago hahaha A Smart UPS with NUT functionality so both UnRaid and the rest of the equipment hooked up to it can shutdown nicely., I Really should start looking for one. Or split the load onto two but we'll see. Thanks again, I'll report back when results are in.CheersAlex
November 6Nov 6 Author Update: In fact the array takes about 4 Minutes to shutdown safely, no wonder I kept getting these unclean shutdowns all the time - the ups is not nearly enough for the load and the time - Well seems like I'm searching for a new one then. In other news: New battery for the current UPS has arrived, will test this today and see if that changes anything.
November 9Nov 9 Author Hello everyone and @JorgeB ;) Quick question: Has anyone experience with the Ubiquiti UPS? Both the 2U and 2U Pro?
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.