To add to my previous post.. Apparently mover has stopped functioning properly as well as my cache drive isn't showing the correct amount of data being used. So that's not good at all. Checking the logs

 

 

Sep 19 08:31:05 Tower nginx: 2017/09/19 08:31:05 [error] 9366#9366: *563455 upstream timed out (110: Connection timed out) while reading response header from upstream, client: 192.168.2.49, server: , request: "POST /update.htm HTTP/1.1", upstream: "http://unix:/var/run/emhttpd.socket/update.htm", host: "tower", referrer: "http://tower/Main"

57 minutes ago, Jerky_san said:

To add to my previous post.. Apparently mover has stopped functioning properly as well as my cache drive isn't showing the correct amount of data being used. So that's not good at all. Checking the logs

 

 

Sep 19 08:31:05 Tower nginx: 2017/09/19 08:31:05 [error] 9366#9366: *563455 upstream timed out (110: Connection timed out) while reading response header from upstream, client: 192.168.2.49, server: , request: "POST /update.htm HTTP/1.1", upstream: "http://unix:/var/run/emhttpd.socket/update.htm", host: "tower", referrer: "http://tower/Main"

Post diagnostics

9 minutes ago, trurl said:

Post diagnostics

Diagnostics option was broke to.. I finally got it to restart and thought it would say it was clean but it wasn't.. Says it came up unclean even though I ran the command via cmd. Couldn't even stop the array. Basically nothing was working. I finally got it back up and just canceled the parity scan since I know it went down clean just doesn't register it as such. If it happens again I'll try to get the diagnostics file before it gets that bad again.

 

Guess reading through didn't realize you can run diagnostics through cmd..

Edited September 19, 20178 yr by Jerky_san

1 hour ago, Jerky_san said:

Diagnostics option was broke to.. I finally got it to restart and thought it would say it was clean but it wasn't.. Says it came up unclean even though I ran the command via cmd. Couldn't even stop the array. Basically nothing was working. I finally got it back up and just canceled the parity scan since I know it went down clean just doesn't register it as such. If it happens again I'll try to get the diagnostics file before it gets that bad again.

 

Guess reading through didn't realize you can run diagnostics through cmd..

 

Are you still having these symptoms?

2 hours ago, Jerky_san said:

Apparently mover has stopped functioning properly as well as my cache drive isn't showing the correct amount of data being used.

Did you set a user share to cache-prefer perhaps? Mover will move any files in a cache-prefer user share TO the cache drive.

11 minutes ago, trurl said:

 

Are you still having these symptoms?

Did you set a user share to cache-prefer perhaps? Mover will move any files in a cache-prefer user share TO the cache drive.

I've not changed anything in unraid for a while now. I generally just follow updates. I was hoping to move to my new ryzen system soon which is the only reason I moved to the RC's.

 

After a restart it takes a day or two for the bugs to come back. It starts with the plugin page's check for updates stops functioning. I usually restart within a day of that happening but I've not because if I restart it will always say it came up unclean now even when using the buttons on the GUI. I've never had the mover/stop array/diagnostic buttons stop functioning until today but its been about a week since the check for updates on the plugin page had stopped functioning. The check for updates on the "Update OS" page stops functioning as well.  Also a first for me to see the cache not updating space utilized. My cache was actually packed full when I restarted. The mover started functioning properly after the restart. 

Edited September 19, 20178 yr by Jerky_san

1 hour ago, Jerky_san said:

if I restart it will always say it came up unclean

Have you tried putting your flash in your PC and letting it checkdisk? unRAID has to be able to write to flash when it shuts down or it will think the shutdown was unclean.

1 hour ago, trurl said:

Have you tried putting your flash in your PC and letting it checkdisk? unRAID has to be able to write to flash when it shuts down or it will think the shutdown was unclean.

I can try and see what happens but it appears that its writing to it. 2017-09-19 14:03 <- last file modified on it so it appears to be functioning properly.

I upgraded my HP Microserver Gen 8, previously running 6.3.5, in order to make use of the 4Kn disk support. Unfortunately my syslog is getting flooded with these ACPI error messages repeated over and over, even before I start the array:

 

Sep 19 21:33:47 Northolt kernel: ACPI Error: SMBus/IPMI/GenericSerialBus write requires Buffer of length 66, found length 32 (20170303/exfield-427)
Sep 19 21:33:47 Northolt kernel: ACPI Error: Method parse/execution failed [\_SB.PMI0._PMM] (Node ffff88020b83c780), AE_AML_BUFFER_LIMIT (20170303/psparse-543)
Sep 19 21:33:47 Northolt kernel: ACPI Exception: AE_AML_BUFFER_LIMIT, Evaluating _PMM (20170303/power_meter-338)
 

Diagnostics attached.

northolt-diagnostics-20170919-2135.zip

 

EDIT: This was my first attempt to run a 6.4.0-rc on this server and the reason for it was specifically to try out the 4Kn support on a WD WD6001FXYZ. I've reverted to 6.3.5 on this server. 6.4.0-rc8q works fine on another server (different hardware). The 4Kn test will have to wait, though I was the one who originally made the feature request and I'm delighted that support has been added.

 

EDIT 2: It seems I'm not alone: https://community.hpe.com/t5/ProLiant-Servers-Netservers/ACPI-Error-SMBus-or-IPMI-write-requires-Buffer-of-length-66/td-p/6943959

Someone suggests a BIOS update but the OP says he's using the latest. I'm not on the latest but I'm out of warranty and haven't been able to find the update.

 

root@Northolt:~# dmidecode -s bios-release-date

07/16/2015

 

A link in the thread sheds a little light. Someone suggests that sensors might be the culprit. I do have the Dynamix System Temperature plugin installed so something I could try is booting into safe mode and if that shows an improvement I could remove the plugin. It only reads the CPU temperature anyway and that isn't very accurate. I'll give rc8q another go tomorrow - I'm currently rebuilding Q parity, having unassigned the original disk in anticipation of putting in the WD6001FXYZ (4Kn) disk.

 

EDIT 3: Ok, the parity rebuild finished so I removed the Dynamix System Temperature plugin and upgraded to rc8q again. The error messages are no longer produced every few seconds but I can still run sensors from the command line and get a meaningful result. Each time I do though, I get the three error messages added to the syslog so I'll take a look at the /etc/sensors3.conf file. But first, I want to try that disk!

 

EDIT 4: The disk was accepted into the array - see my later post. Switching to the Dashboard in the GUI presumably calls sensors because it causes the three ACPI error messages to to added to the syslog.

 

EDIT 5: I think I see what the ACPI problem is. See this bug report.

 

root@Northolt:~# sensors

coretemp-isa-0000

Adapter: ISA adapter

Package id 0:  +37.0°C  (high = +87.0°C, crit = +91.0°C)

Core 0:        +37.0°C  (high = +87.0°C, crit = +91.0°C)

Core 1:        +36.0°C  (high = +87.0°C, crit = +91.0°C)

 

acpitz-virtual-0

Adapter: Virtual device

temp1:         +8.3°C  (crit = +31.3°C)

 

power_meter-acpi-0

Adapter: ACPI interface

power1:        0.00 W  (interval = 300.00 s)

 

The power_meter-acpi-0 sensor seems to be the culprit. The fix seems to be to edit the config file to ignore the power meter sensor. But how to do that and make it permanent?

 

Edited September 20, 20178 yr by John_M
More information

On 9/17/2017 at 8:03 PM, ljm42 said:

On the other hand, if you go to:
  http://tower
there is no encryption involved, and unRAID will be able to redirect you to the FQDN:
  https://6c4044c77838a17a5d5c58ddb2e63cbb7ad07b4d.unraid.net

 

This will work if USE_SSL is set to Auto or Yes.

 

If it's set to Only, there's no redirection, so accessing http://tower will fail 

And another crash.... this is gettign somewhat annoying.. Has anyone had a chance to check out my logs ?

 

Last comment in the log:

 

Sep 19 23:20:01 Tower kernel: Call Trace:
Sep 19 23:20:01 Tower kernel: out_of_memory+0x3ac/0x3ce
Sep 19 23:20:01 Tower kernel: Killed process 14830 (shfs) total-vm:25346876kB, anon-rss:22499160kB, file-rss:4kB, shmem-rss:808kB

 

tower-diagnostics-20170920-0725.zip

Edited September 20, 20178 yr by Helmonder

@bonienl

I did a search and couldn't find it reported. It is the first time I have used the plugin, but system temps dont show up for each tab, e.g. when clicking on dashboard or main, system temps disappear. Same with the docker / VM tab.

 

 

Edited September 20, 20178 yr by Lebowski

9 minutes ago, Lebowski said:

@bonienl

I did a search and couldn't find it reported. It is the first time I have used the plugin, but system temps dont show up for each tab, e.g. when clicking on dashboard or main, system temps disappear. Same with the docker / VM tab.

 

 

Ignore me, seems it can take 9-10 seconds to show up on the dashboard.

 

However it instantly shows on the shares & settings TAB. Slightly strange.

Edited September 20, 20178 yr by Lebowski

6 hours ago, Helmonder said:

And another crash.... this is gettign somewhat annoying.. Has anyone had a chance to check out my logs ?

 

Last comment in the log:

 

Sep 19 23:20:01 Tower kernel: Call Trace:
Sep 19 23:20:01 Tower kernel: out_of_memory+0x3ac/0x3ce
Sep 19 23:20:01 Tower kernel: Killed process 14830 (shfs) total-vm:25346876kB, anon-rss:22499160kB, file-rss:4kB, shmem-rss:808kB

 

tower-diagnostics-20170920-0725.zip

 

I haven't looked for your earlier posts but have you tried this yet?

It seems to have worked for a lot of folks...

Confirming that my 4Kn disk was accepted into the array as Parity2. Many thanks for that. It took a lot longer than I'm used to for the array to start - not a problem in itself, but my palms began to sweat! Q parity is now rebuilding. There's a slight bug in the display of the write speed of the 4Kn disk (I don't know whether read is also affected) - it seems to be out by a factor of 4096/512 = 8:

 

 

5 hours ago, Frank1940 said:

 

I haven't looked for your earlier posts but have you tried this yet?

It seems to have worked for a lot of folks...

 

Thanks but this really appeared in the latest unraid version.. So something is going wrong there.. Maybe it can be circumvented with this though..

33 minutes ago, John_M said:

There's a slight bug in the display of the write speed of the 4Kn disk (I don't know whether read is also affected) - it seems to be out by a factor of 4096/512 = 8:

 

Can you show the output of:

 

cat /sys/block/sdc/queue/hw_sector_size

and

cat /sys/block/sdc/queue/logical_block_size

 

3 minutes ago, bonienl said:

 

Can you show the output of:

 

cat /sys/block/sdc/queue/hw_sector_size

and

cat /sys/block/sdc/queue/logical_block_size

 

 

root@Northolt:~# cat /sys/block/sdc/queue/hw_sector_size

4096

root@Northolt:~# cat /sys/block/sdc/queue/logical_block_size

4096

 

3 minutes ago, Helmonder said:

 

Thanks but this really appeared in the latest unraid version.. So something is going wrong there.. Maybe it can be circumvented with this though..

 

It just suddenly popped up as an issue to most of the folks who have encountered it.  It is quick and easy to try without any real side effects...

4 minutes ago, John_M said:

 

root@Northolt:~# cat /sys/block/sdc/queue/hw_sector_size

4096

root@Northolt:~# cat /sys/block/sdc/queue/logical_block_size

4096

 

Thanks for the quick reply, one more!

 

cat /sys/block/sdc/queue/physical_block_size

 

6 minutes ago, bonienl said:

Thanks for the quick reply, one more!

 

cat /sys/block/sdc/queue/physical_block_size

 

No problem. Thank you!

 

root@Northolt:~# cat /sys/block/sdc/queue/physical_block_size

4096

 

I fixed my ACPI issue (here) by adding

 

chip "power_meter-*"

    ignore power1

 

to the end of the /etc/sensors3.conf file, as recommended in this bug report. Now I can run sensors from the command line and I can switch to the Dashboard page of the GUI without getting the error messages in my syslog. I didn't need to re-load anything - just edit the file and save it.

 

I suppose I ought to have left the file alone and put something in /etc/sensors.d instead, as recommended at the beginning of the sensors3.conf file. In any case it won't survive a reboot but it proves the idea. Could this little tweak be incorporated into the next rc, please?

 

I've now re-installed the Dynamix System Temperature plugin and, apart from the mentioned ACPI error messages appearing when the plugin was first probing for sensors, it is displaying the CPU temperature continuously without causing any more syslog entries. I notice the plugin creates a file called /etc/sensors.d/sensors.conf so perhaps that's where the tweak to ignore the power meter sensor ought to go.

 

And the third crash...

 

 

 

6 minutes ago, Helmonder said:

And the third crash...

 

Can you try 6.4.0-rc9f ?

 

And the third crash...

 

 

 

 

Looks like it's always java running out of memory, maybe a plugin?