[Sata/SAS controllers tested (real world max throughput during parity check)]

On 9/21/2015 at 4:47 PM, JorgeB said:

*Despite being DMI 3.0** , Skylake, Kaby Lake, Coffee Lake, Comet Lake and Alder Lake chipsets have a max combined bandwidth of approximately 2GB/s for the onboard SATA ports.

How did you get that information? What about a W480 chipset with 8 SATA ports? (Gigabyte W480M Vision W)

[ASM1166 brakes ASPM]

On 11/20/2023 at 5:18 PM, etnicor said:

Bought an ASM1166 card from amazon which should be gen3 x4.

 

From what I can read it should asm1166 is x2 not x4...

 

My problem is that the card brakes cpu package c-state and cpu can only reach c3 instead of c6-c7. Is this an issue with asm1166 or just my card? lpsci show that ASPM is enabled, so no idea why it block cpu states from going lower. I am guessing firmware/driver issue?

 

 

07:00.0 SATA controller: ASMedia Technology Inc. ASM1166 Serial ATA Controller (rev 02) (prog-if 01 [AHCI 1.0])
                LnkCap: Port #0, Speed 8GT/s, Width x2, ASPM L0s L1, Exit Latency L0s <4us, L1 <64us
                LnkCtl: ASPM L1 Enabled; RCB 64 bytes, Disabled- CommClk+

 

 

EDIT: Switching to kernel  6.5.11-3(not unraid) seem to have solved the issue for me.

how did you switch kernels, if you don't mind me asking? did you just test with a different OS?

[Unassigned Devices - Managing Disk Drives and Remote Shares Outside of The Unraid Array]

thank you

1 hour ago, dlandon said:

Go to the UD page and click on 'Help' in he upper right hand corner of the page and read down about the rc.unassigned script.

 

[Unassigned Devices - Managing Disk Drives and Remote Shares Outside of The Unraid Array]

hi,

is there a way to call the mount of a remote share setup via unassigned devices if it get's unmounted without restarting the array (for automount) or manually clicking the mount button? I am looking to mount my backup nas (which only gets powered on once a week for backups) within unraid. It is configured via unassigned devices but gets taken offline, when the backup nas poweres off and not remounted, when the backup nas comes online again.

 

thank you!

[periodic system lock up/freeze]

i did... and i am running a seperate syslog server... but nothing in the logs... like nothing. just being informed of the occasional spin down or write method switch.

will post after the next crash

[periodic system lock up/freeze]

hi,

 

i seem to be in the same boat as some other users where unraid locks up completely from time to time. i did check my powertop settings and disabled all power saving for the connected drives and the asm sata controller.

when unraid freezes i can not login via web/ssh/console.

 

system:

xeon w-1290p

Gigabyte W480 Vision D

32gb of ECC Kingston RAM

3x 18TB Toshiba

2x 960GB Samsung SSD

2x 3.84TB Samsung SSD

2x 1TB Samsung 980 NVME SSD

1x ASM1166 with updated Silverstone Firmware for ASPM compatibility

10G AQC107 NIC

 

any help possible? it bugs me to hell

jarvis-diagnostics-20231006-1134.zip

[server lock up - now installed plugin tab is empty]

i solved this in the mean time. a replicated job for some zfs datasets with mountpoints set messed up some of unraids mount points.

[server lock up - now installed plugin tab is empty]

so my server locked up a few days ago (nothing in the logs - i suspect powertop messing up the wrong device). Anyway, since the reboot the installed plugin tab shows no installed plugins (they show up in settings and work, by all accounts). Also on the update os tab, i can only see the restore option and, funny enough, a message that no plugins are installed.

 

is there a way to reset the webui or something without reinstalling everything (i have backups via myserver and other means)

[what about array drives with ZFS filling up]

i was thinking about using my new 18tb mg09 as array drives in zfs (for easy zfs send/receive and compression). what happens to my array if one of the drives filles up to maybe 50GB free space? i know performance is going to tank hard on the way there but will it render the filesystem (or array) unusable? since this never happened to me before i could use some input.

 

thank you!

[[Support] ich777 - AMD Vendor Reset, CoralTPU, hpsahba,...]

On 7/8/2023 at 7:08 PM, ich777 said:

I can pass that through to Tom and see if he implements those modules directly into Unraid.

 

But clockspeeds (prometheus_node_exporter) and also the power draw (intel_gpu_top) should be visible already or am I wrong?

 

EDIT: I did a bit of digging:
X86_MSR is already available

CONFIG_INTEL_RAPL is not enabled and I'll have to look into what it depends on but I think on nothing else then the third listed and vice versa
CONFIG_INTEL_RAPL_CORE is also not enabled and the same as for the second listed applies

i can use most of the functionality of the powerstat plugin for telegraf (package power) - but frequency and c-state are not working yet. i suppose that's what intel_rapl_core is for.

any news on if it's possible to implement?

 

THANK YOU for the great support!

[idle power draw on xeon w-1350]

hello,

i was using a i7-10700 until a short while ago. i got it down to about 4-5 w idle (on the cpu) with my dockers running etc. . since i switched to a xeon w-1350 the idle power draw has doubled. powertop --auto-tune, powersave cpu governor and bios settings limiting pl1 and pl2 were all tried.

 

any pointers? are xeon w just not able to idle that low?

 

edit: i am using the intel.powerstat plugin within telegraf to monitor cpu draw (THANKS @ich777, btw it's working) - but my external meter confirms the extra 5-6 w idle load

[[Plugin] LXC Plugin]

is there a way to limit cpu usage or define cores to be used? ram limit?

normal lxc commands obviously dont work here.

 

nevermind, found the post in this topic... (search DOES work)

[[Support] ich777 - AMD Vendor Reset, CoralTPU, hpsahba,...]

4 hours ago, ich777 said:

I can pass that through to Tom and see if he implements those modules directly into Unraid.

 

But clockspeeds (prometheus_node_exporter) and also the power draw (intel_gpu_top) should be visible already or am I wrong?

 

EDIT: I did a bit of digging:
X86_MSR is already available

CONFIG_INTEL_RAPL is not enabled and I'll have to look into what it depends on but I think on nothing else then the third listed and vice versa
CONFIG_INTEL_RAPL_CORE is also not enabled and the same as for the second listed applies

thank you! sounds promising.

[[Support] ich777 - AMD Vendor Reset, CoralTPU, hpsahba,...]

@ich777Hi,

i was wondering how to add specific kernel modules to unraid (rapl, msr, intel_rapl_common and intel_rapl_msr) for the telegraf intel.powerstat plugin, so i can monitor power draw and clock speeds in grafana. After googling for a bit i found your unraid-kernel-helper but it seems to be way out of date and not "supported" anymore.

 

Is there still a way to load those modules? can i use the latest version of the plugin still - can't find any .plg for it anymore.

 

thank you for your great work!

[auto-completion in cli]

hi, is there a way to get command auto-completion (on tab press) for the cli?

thank you for your consideration

[Dynamix File Integrity plugin]

On 6/14/2022 at 10:35 AM, santinilor said:

Hi,

 

I've installed the File Integrity plugin and successfully run the first build. I activated the option "Automatically protect new and modified files" but it appears to me that new hashes aren't created automatically. I tried to add a test file to an user share and still, after a couple of days, if I check the exported file (after a new export) the test file hash doesn't appear to be present. Furthermore on the File Integrity control page the disk with the file shows an orange circle near "build up to date". If a do a manual build the circle disappears and the new file hash is created but the whole point of this plugin for me is to automatically protect new files and not have to build it manually each time.

 

 

 

Hope someone can help me.

 

Here I attach the diagnostic file

jarvis-diagnostics-20220614-1033.zip 183.3 kB · 0 downloads

same for me - have been struggling with this for a while. all data is coming in via smb or docker bind mounts. does it only work on files changed via smb?

[slow smb transfer/single connection limited to 800mbit/s]

hi,
i have unraid running as a guest in proxmox with a virtio network adapter.
running iperf i get about 30 gbit/s from host to guest.

guest to host is limited to 700-900 mbit/s per connection with a multi-gbit peak at the start.

running multiple threads i can get up to 30-40 gbit/s. with single connections running ABOVE 1 gbit/s as desired.

there is no tuning happening on the unraid side and cpu usage on host and guest is low (no cores/threads pegged) during the the tests.

after disabling the network bridge in unraid i can get good bandwidth (iperf) between host and guest in either direction... now its smb that is limited to ~700-800 mbit/s from guest to host. HELP?

 

EDIT: transfer from host to guest is going full speed. only from guest to host is limited somehow.

[[Plugin] Mover Tuning]

is there a way to make the mover run every 2 days. does using the cron schedule in the plugin force the "normal" mover to run every 2 days instead of every day?

[Dynamix - V6 Plugins]

hi,

i find that after restarting my server there seems to be some (? run-away ?) sh process linked to folder caching. it easts cpu cycles in regular peaks and does not ever stop. i have folder caching enabled (but not for /mnt/user, because that eats even more power). killing the process gets the cpu usage down to 'normal' idle levels. it reappears after reboot though. any tips on what i am seeing here?

[Recycle Bin (vfs recycle) for SMB Shares]

7 hours ago, dlandon said:

It mounts either /mnt/user/ or /mnt/user0/ (selectable) and shares that mount and applies the recycle bin parameters to the rootshare.  All the 'shares' work the same as they do separately.

 

The warniing applies to the UD rootshare and is the same if you use the smb-extra.conf configuration.  It's the way files are allocated to shares.

Thank you for the explanation. Just one more thing: Is there a way to control access to the ud rootshare? like user-management?

[Recycle Bin (vfs recycle) for SMB Shares]

then i misread the situation. i thought because of the way it displays, that it mounts a share and then shares it. the warning says it can not be guaranteed that files end up on the correct drives. is that an edge case or something that can be avoided?

thanks for all the info

[Recycle Bin (vfs recycle) for SMB Shares]

right now i am using a symlink from one of the cache drives (a seperate one just for appdata). it points to /mnt/user and works well enough (even with recyclebin). the only downside i can see so far is a folder called 'root' (name of the symlink/share) that sits in the rootshare and points in a loop to the same folder :D. i set the cache policy for it to no and it falls back to the individual cache settings of the underlying shares respectively. is the performance any different when using the ud rootshare, since it chains cifs mounts?

[Recycle Bin (vfs recycle) for SMB Shares]

Hi,

can someone confirm, that the recycle bin is now working with the rootshare setting? and if so which rootshare type do we have to use (UD vs samba extra config)?

 

thanks for you work!

[drive in parity is not the biggest]

i see. the drives were carried over from another nas setup - i had unraid format them. should i use pre-clear next time or does deleting the old partition before formatting them next time?

[drive in parity is not the biggest]

yes you are correct, which is still a strange thing because i've created and formatted the drives using unraid. why would it put the partitions down inconsistently?