October 3, 200817 yr @NAS, I started writing a process to do this, using hdparm -Y to force the drive down, and not allow it back up without a reboot (or a hdparm -w which is fraught with peril). I ran into this issue: If you don't stop the array first, it flags the disk bad, and you have to have a parity check on startup. I was playing around with removing the SAMBA share for that drive and unmounting it, so that no writes could be attempted to it, so I could force parity good on reboot, but I got distracted by the discovery that WD Green drives return temp w/o spinup, and changed the code to kick the fans on the drive cages into turbo when a high-temp is detected, rather than sleep the drive.
October 3, 200817 yr Thats a side effect no one could live with. I cant think of another solution but surely there must be one?
October 3, 200817 yr Thats a side effect no one could live with. I cant think of another solution but surely there must be one? Forced shutdown would only work with solid notification - preferably with a different color "dot" on the main screen. If heat were to cause the entire unraid to be shut down, then we need some other form of notification - email preferably. Bill
October 3, 200817 yr Well, since an overtemp drive is potentially a VERY serious problem, a shutdown of the drive and subsequent parity rebuild may be an appropriate reaction. hdparm -C will report "standby" for a spun-down drive, but if you shut a drive down with -Y so if can not be spun back up w/o a reboot, hdparm -C will report "sleeping".... so unRAID could display it differently in the interface. Obviously, you have to keep track and if a second drive reaches shutdown temp, the array MUST be shut down. Writing to the shut-down drive is the pig... a write will change the parity drive, but not be written to the disk, so you MUST rebuild the drive when the system restarts.
October 3, 200817 yr I recommend shutting down the entire array on overheat of any one drive. If a drive is overheating, there is a serious issue with the system -- and trying to limp along by shutting down only the offending drive (placing it in a failed state) would likely be counter productive.
October 3, 200817 yr There should be plenty of time to complete the write actions before shutting down the array. I would suggest setting the high-temp at something like 50C, low enough to give that time to finish the writes but high enough to avoid false positives. Bill
October 3, 200817 yr My small contribution. Personally, I would stop the array and spin down all the drives. Add something that highlights the problem on the main page. I would put a drop down from say 40C to 60C for over temperature or a box you can fill in yourself with limits from 30C to 70C or something similar. Everyone has their own drive temp preferences. Peter
October 3, 200817 yr Is there a way to get the sata_nv.swncq=0 parameter to the sata_nv module without rebuilding the initrd filesystem or the kernel? Seems like the kernel rejects it as a option coming in from syslinux, unless sata_nv was built-in rather than a module - or so I infer from what I've read online. Sounds like this is whats causing my 4.4 woes as well (Thanks RobJ for the info)
October 4, 200817 yr Is there a way to get the sata_nv.swncq=0 parameter to the sata_nv module without rebuilding the initrd filesystem or the kernel? Seems like the kernel rejects it as a option coming in from syslinux, unless sata_nv was built-in rather than a module - or so I infer from what I've read online. Sounds like this is whats causing my 4.4 woes as well (Thanks RobJ for the info) It did not work for me either, syslog says: Unknown boot option `sata_nv.swncq=0': ignoring I then decided to try just swncq=0 instead, and it *seems* to be working much better, so far. There is no error in the syslog, so it accepted it. And my acid tests of clicking the Spin Up button and the Stop button have produced NO errors! Those 2 buttons and any other action that causes multiple drives to spin up have been creating multiple exception Emask error sequences, once disabling one drive, with any of the v4.4 betas up 'til now. Thankfully, when the drive was disabled, Tom's procedure (explained here) came to the rescue. The one thing that still concerns me is that the sata_nv module is still using SWNCQ mode: sata_nv 0000:00:05.0: Using SWNCQ mode. Clearly, SWNCQ must apply to more than sata_nv. I'll let you know if I start seeing more errors, only rebooted with it within the last hour, but so far so good.
October 4, 200817 yr I then decided to try just swncq=0 instead, and it *seems* to be working much better, so far... Well, the kernel didn't complain about that option, but no change in behavior for me - my WD5000AAKS drives still won't come up. I also tried putting "option sata_nv swncq=0" in a modprobe.d/ file with the same results. Maybe I'll try building 2.6.27-rc8 and see what happens.
October 4, 200817 yr I probably won't be able to help you, but I'm curious what you see in the syslog, when you say they don't come up. Have you compared a v4.3 syslog with a v4.4 syslog? Is it the controller or port not found, or the drive not found on that port (should say 'SATA link down'), or it finds the drive but disables it with errors?
October 5, 200817 yr Tom, is there any way the Cache drive portion could be mod'd to allow mounting of a partition on a drive as cache instead of requiring the entire drive? Do you mean a separate partition on one of your data drives? If so, no. But, you can use the 'fdisk' command to manually create multiple partitions on your cache drive and unRAID will use partition 1 (so you have to have a partition 1). But you could then use other partitions on that drive for non-unRAID purposes. If you want to do this though, you have to also manually create the ReiserFS file system in partition 1 as well before assigning the drive as the Cache drive to prevent System Management utility from wanting to format it (which will also create a new partition table). Note this will work only with the Cache drive, not the parity/data drives. Make sense? okay, I've got this working now but, as usual, unintended consequences... can we have a radio button or something on the settings page to disable the cache drive from the "spindown" button on the main page? Also, if a process outside the array is using another partition on the "cache drive", is unraid smart enough to not automatically spin it down?
October 5, 200817 yr Author okay, I've got this working now but, as usual, unintended consequences... can we have a radio button or something on the settings page to disable the cache drive from the "spindown" button on the main page? Also, if a process outside the array is using another partition on the "cache drive", is unraid smart enough to not automatically spin it down? We're adding separate spin-down controls for each drive, so you will be able to select 'Never' for the cache drive if you wanted. The unRAID spin down logic looks at I/O statistics for each drive to determine how long a disk has been idle. These statistics are independent of which process is used to access the drive, and which partition is being accessed. So answer is yes
October 10, 200817 yr Can someone help me understand the FTP functionality in this release. I have user shares that are populated via rSync (backup), FTP (network cameras), and I'm also uploading music & movies over my network by mapping the shares onto other machines. So, my files are finding their way to my unRAID server several ways. The problem I have is when I connect via FTP to browse the contents of my server. I can see all the disks & user shares, and all the subdirectories that they contain. However, the only files I can see in those disks/shares/directories are those that found their way there via rSync. Could this be a permission issue? I tried playing with file permissions using CHMOD, but that didn't seem to make any difference.
October 10, 200817 yr Could this be a permission issue? I tried playing with file permissions using CHMOD, but that didn't seem to make any difference. Hi, I've had the same issues. Have a look here, a few posts down: http://lime-technology.com/forum/index.php?topic=2171.0 Cheers Søren
October 10, 200817 yr Søren, Thanks. I think you might be correct. I learned a lot from reading that thread (I've got more to learn about file and directory permissions but at least I have some clue to start with now.) I tried this command: find /mnt/disk2 -type d -exec chmod g+rx,o+rx {} \;[/b] But, I get an error "find: missing argument to `-exec'. Did you have this problem too?
October 10, 200817 yr Søren, Thanks. I think you might be correct. I learned a lot from reading that thread (I've got more to learn about file and directory permissions but at least I have some clue to start with now.) I tried this command: find /mnt/disk2 -type d -exec chmod g+rx,o+rx {} \;[/b] But, I get an error "find: missing argument to `-exec'. Did you have this problem too? You copied too much from a prior post. The correct command is: find /mnt/disk2 -type d -exec chmod g+rx,o+rx {} \; There should not be a "[ /b ]" at the end of the command.
October 10, 200817 yr Today the kernel numbered 2.6.27 got relased. Will you update it in the beta3 relase and implement full NFS support, as you mentioned Tom?
October 10, 200817 yr Today the kernel numbered 2.6.27 got relased. Will you update it in the beta3 relase and implement full NFS support, as you mentioned Tom? you probably did not see this post: http://lime-technology.com/forum/index.php?topic=2575.msg20720#msg20720
October 12, 200817 yr Author Today the kernel numbered 2.6.27 got relased. Will you update it in the beta3 relase and implement full NFS support, as you mentioned Tom? That will be the first thing we address as soon as we move to the 2.6.27 kernel.
October 23, 200817 yr Tom hopefully you will see this. Please please please please can we have wget in the next release... please
October 23, 200817 yr I'm going to go out on a limb and suggest curl. I happen to use wget all the time, but I think curl is more comprehensive. it handles all the protocols that wget does, but it also does pushes. This is sometimes a big hindrance when I'm testing new hardware with unRAID. I boot up, it's failing, the network has not been allocated and after I allocate it, I can not push a file out. We don't have a basic FTP client, With curl a file can be pushed like curl -T $file ftp://uploadsite.com/directory/$file Furthermore it supports mutilpart form/file uplodas, this could provide some interesting additions in the future, such as automated syslog uploading. Automated product downloads for updates. In addition, the curl package would contain the libraries that many other addons would use. (I.E. php and others) Related links. http://daniel.haxx.se/docs/curl-vs-wget.html http://curl.netmirror.org/docs/httpscripting.html http://en.wikipedia.org/wiki/CURL
Archived
This topic is now archived and is closed to further replies.