Jomp Posted April 7, 2014 Share Posted April 7, 2014 I have a weird one. During my build (http://lime-technology.com/forum/index.php?topic=32647.15), everything went great on my first parity sync that I ran overnight. The next day, I decided to run another one and I got SATA link reset errors. I changed cables and they were still occurring. So through a process of elimination and a little but of luck, I narrowed it down to the webgui, of all things. So if I start a parity sync and leave everything alone, it will complete without any SATA link resets. If I start a parity sync and pull up the webgui, and click between the tabs on the menu bar (Main, Users, Settings, Utilities) it will eventually freeze the webgui for ~10 seconds and the log will show a SATA hard link reset. Sometimes it takes ~3 tab clicks, sometimes up to ~20. Weird. What in the heck could be going on? Edit: It happens on 5.0.5 and 6.0beta 4 log4.txt Link to comment
SSD Posted April 7, 2014 Share Posted April 7, 2014 Don't have a good answer, but the Web GUI is querying the drives smart attributes to get temperature info. I can say that the worst problem drive I ever had was an older seagate that did something similar during preclears. Check the smart reports carefully. You might try a different controller. Link to comment
Jomp Posted April 7, 2014 Author Share Posted April 7, 2014 To add some more datapoints: It happens during a parity check as well. Pretty much any parity operation with reads or writes combined with menu activity causes sata resets. I tried Dynamix and it freezes sata links on every page load versus the more intermittent limetech webgui. Is it something with the SMART queries during high use operations? These are the controllers giving me issues: Marvell SE9172: 2 x SATA3 6.0 Gb/s - Marvell SE9230: 4 x SATA3 6.0 Gb/s I'll try menu presses during a protected write next to see if a less intense I/O operation causes the sata resets. Link to comment
dmacias Posted April 7, 2014 Share Posted April 7, 2014 When you're running 6b4 do you rename /plugins or the webgui.plg? I wonder if it could be a driver or kernel issue too or something with NCQ implementation on the Marvels. Have you tried force NCQ disabled? Link to comment
Jomp Posted April 7, 2014 Author Share Posted April 7, 2014 When I transfer a file to the protected array and use the menu at the same time, I'm getting sata resets. So there we have it, any disk activity and the use of webgui causes sata link resets. Link to comment
dmacias Posted April 7, 2014 Share Posted April 7, 2014 Is it still the same drives or all on the marvel? Did you see above about ncq? Link to comment
Jomp Posted April 7, 2014 Author Share Posted April 7, 2014 Is it still the same drives or all on the marvel? Did you see above about ncq? All the drives are on the Marvels. I can't find many Marvel sata settings. You can enable/disable SMART, set sata/IDE mode on two of the ports and that's about it. The Intel sata has far more options. Link to comment
dmacias Posted April 7, 2014 Share Posted April 7, 2014 The 2 areas for storage config in the bios are under Advance/Storage Configuration and under IntelRCSetup/Southbridge Chipset Configuration/Sata Configuration don't forget to page down also for more. This is where the aggressive link is The force NCQ disabled is under the unraid webui in disk settings. I would try the bios settings first then ncq also go into each through each drive in the bios and turn off hot swap Edit: Is the marvell just the first page and the second is all Intel? Edit2: I seem to remember there being a hot key during POST for the Marvel at some point. Maybe I'm wrong or maybe it had to be set to bootable in the bios. I'll check some stuff later when I change the power supply. Link to comment
Jomp Posted April 7, 2014 Author Share Posted April 7, 2014 The 2 areas for storage config in the bios are under Advance/Storage Configuration and under IntelRCSetup/Southbridge Chipset Configuration/Sata Configuration don't forget to page down also for more. This is where the aggressive link is The force NCQ disabled is under the unraid webui in disk settings. I would try the bios settings first then ncq also go into each through each drive in the bios and turn off hot swap Edit: Is the marvell just the first page and the second is all Intel? Edit2: I seem to remember there being a hot key during POST for the Marvel at some point. Maybe I'm wrong or maybe it had to be set to bootable in the bios. I'll check some stuff later when I change the power supply. Force NCQ disable did not fix it. All the Marvell settings that I could find were under the Advanced tab. I can't find a way to edit hot swap or aggressive link on the marvells, just on the intel controller under the southbeidge menu. Link to comment
dmacias Posted April 8, 2014 Share Posted April 8, 2014 When I installed my new power supply today I enabled boot for the marvel and pressed ctrl-m during boot but there isn't anything in there but raid setup. Only other things I can think of is to move your parity to the Intel controller and turn off ecc or try some non ecc memory. Link to comment
Jomp Posted April 8, 2014 Author Share Posted April 8, 2014 I installed FreeNAS (FreeBSD) and when writing to the array and issuing SMART commands, I got SMART timeouts. So there appears to be some problems with the Marvell controllers on the C2550D4I. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.