crankbearing Posted April 29, 2012 Share Posted April 29, 2012 Tom, Just updated to rc1 from b14 it fixed the acpi namspace error still is a warning but not in red. But I am still getting ata9 and ata10 sas errors in the first boot up. The other thing is the system came back up with the array already started. syslog attached. I posted earlier about lm-sensors, i guess it got missed with the larger issues at hand, but is it included and what is the version please. syslog-2012-04-29.txt Quote Link to comment
jumperalex Posted April 29, 2012 Share Posted April 29, 2012 I'll just quote myself from earlier as nobody answered my question: OK I have some questions. 1: Can someone provide a list of current known bugs that 5.0-RC1 does NOT address/fix. As far as I am aware the only known bugs are: - LSI Controller card support - Major drop in parity rebuild speed reported since beta 12 ? With regard to LSI supported cards, I just had a quick look on a well known UK supplier of computer hardware, and 8 port cards are available from 5 manufacturers: Areca Highpoint LSI Lycom Supermicro Which of the above would be based on the LSI chipset, or would it depend on a particular card? -beta12 fixed a parity sync speed issue. I see no speed decrease on any of my test systems. FWIW, the below specs just got me 75MB/s on a pre-RC1 upgrade parity check. unRAID Version: unRAID Server Plus, Version 5.0-beta14 Motherboard: ASUSTeK - M4A785-M Processor: AMD AthlonTM 7750 Dual-Core - 2.7 GHz Cache: L1 = 256 kB L2 = 1024 kB L3 = 2048 kB Memory: 4608 MB - DIMM0 = 800 MHz DIMM1 = 800 MHz Disks: 3x Hitachi 2TB connected to onboard SATA filled with 1.2TB Quote Link to comment
Marky Posted April 29, 2012 Share Posted April 29, 2012 So finally taken the plunge and upgraded from 4.7 to 5.0RC1. Upgrade went fine but i am also seeing parity speed problems. On 4.7 i was getting ~55MB/sec On 5.0RC1 i'm getting ~17.50MB/sec This is a huge speed drop. This is on a Supermicro C2SEA board as recommended by Limetech so would have thought Tom you would have seen it on one of your test systems. Where are you located in Carlsbad Tom? I live and work there so if a meetup is needed to find out whats going on here it might be possible. Mark Quote Link to comment
Marky Posted April 29, 2012 Share Posted April 29, 2012 Hmmm so now after an hour the web interface has died. Can telnet in ok so the machine has not crashed. Is there a way to stop the parity check from the command promt and do a safe shutdown? Was only doing a check with no correct just as a verification after the upgrade. Thanks Mark Quote Link to comment
EasyME Posted April 29, 2012 Share Posted April 29, 2012 I've seen it where the web interface has become unresponsive when doing parity check (both correct and none correct checking), I would say let it go to complete the parity check and then check the web interface after this is done. It has come back / responsive after this. Quote Link to comment
Marky Posted April 29, 2012 Share Posted April 29, 2012 Funnily enough the web interface has started responding again while the parity check is in progress. The parity speeds still concern me though its now so slow. Tom any ideas or any tests you would like me to do? Thanks Mark Quote Link to comment
burnaby_boy Posted April 29, 2012 Share Posted April 29, 2012 After the upgrade to rc1 from beta 14 I've noticed a significant decrease in speed when copying to the server (with parity) - from 27 MB/s with beta 14, down to 10 MB/s with rc1. I'm using a Gigabyte P35-DS3P with a Supermicro AOC-SASLP-MV8 and an Adaptec 1430SA. The syslog is attached. Looks like I'll head back to beta 14. Update - after copying the beta 14 files back on to the flash drive and rebooting, the web interface eventually reappeared, but with a parity check already in progress. syslog.txt Quote Link to comment
brent112 Posted April 29, 2012 Share Posted April 29, 2012 I've seen it where the web interface has become unresponsive when doing parity check (both correct and none correct checking), I would say let it go to complete the parity check and then check the web interface after this is done. It has come back / responsive after this. I have seen the same thing. If i do a "mdcmd nocheck" to kill the parity check the web interface comes back. not sure where the problem lies. Quote Link to comment
Joe L. Posted April 29, 2012 Share Posted April 29, 2012 Hmmm so now after an hour the web interface has died. Can telnet in ok so the machine has not crashed. Is there a way to stop the parity check from the command promt and do a safe shutdown? Was only doing a check with no correct just as a verification after the upgrade. Thanks Mark I would do a tail -f /var/log/syslog to see if you are getting disk errors which will have filled your system log and will eventually cause an out-of-memory failure. Quote Link to comment
_vern_ Posted April 29, 2012 Share Posted April 29, 2012 I have been running 5.0-beta13 for what seems to be forever now with no issues. I normally get copy speeds of 20 to 25 MB/s from my torrent box to my unraid box, no cache drive. I rolled the dice and went ahead and upgraded to rc1 and now I'm lucky to hit 9 MB/s and there is considerable stuttering during the copy operation. I don't run any add-ons and haven't changed anything other than copy over the bzimage and bzroot files. I have rebooted a couple of times just to make sure and I get the same behavior. I looked through syslog and nothing jumps out at me. The network interface still reports "eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX/TX" and I haven't changed anything with the network, so it makes me wonder if something is wrong with SMB being that is one the things mentioned in the change log. I have attached my syslog for review. If anyone has any thoughts on it, I'd sure appreciate it. Thanx syslog_4_29_2012_1420.txt Quote Link to comment
Marky Posted April 29, 2012 Share Posted April 29, 2012 So a little more info on parity speeds. Like i posted earlier. 4.7 was ~55MB/sec 5.0RC1 was ~17.50MB/sec I've now dropped back to 5.0B14. Now parity check is running at ~50MB/sec. So RC1 certainly has some issues. Quote Link to comment
Fred Posted April 29, 2012 Share Posted April 29, 2012 Getting 53.8mb/sec parity sync on this release. One issue is I can't get "flash" share to actualy export syslog_fred.txt Quote Link to comment
_vern_ Posted April 29, 2012 Share Posted April 29, 2012 FYI...I downgraded to 5.0-beta14 and I'm back to ~20-25 MB/s transfers. I guess I'll just stay here until I see something different...hence beta... Thanx Quote Link to comment
jumperalex Posted April 29, 2012 Share Posted April 29, 2012 And for more reference to my previous B14 post ... now I'm seeing RC1 parity speeds on the same order as Beta 14 ... so at least for my config (read: onboard SATA slots no add-on cards) Quote Link to comment
limetech Posted April 29, 2012 Author Share Posted April 29, 2012 Here is the list of issues gleened from this thread: -several users: lsi based controllers have issue accessing spun down disks -sacretagent: wants 24 disk support -kingping: reports "hdparm sending ioctl 2285 to a partition!" filling system log -crankbearing: wants to know what version of lm_sensors is being used [it's 3.1.2] -peter_sm: has issue where subtitles disappear using NFS -dlandon: clicking utils/system log constantly refreshes using IE9 -darkside40: samba 3.6.3 has a vulnerabilty which apparently has a patch somewhere -marky (and others): webGui freezing, especially during parity sync -burnaby_boy, _vern_: writes via network much slower vs. -beta14 -Fred: can't get flash share to export It looks like the last kernel that was reliable with the LSI based controllers was the 3.0.x series. It also appears the 3.0.x series is being actively maintained. So what I'm going to do is take down -rc1 and rebuild with latest 3.0 kernel (3.0.30) and release as -rc2. Sorry -rc1 has turned out to be such a lemon. Some things are my fault, some are issues in the linux kernel. As an aside - I liked linux kernel development method the way it used to be: if the second digit was even, it meant it was a "stable" release, if odd it meant it was a development release. Now it's hard to know when to update the kernel. I will also say this: not everyone is experiencing the kinds of issues being reported in this thread, and none of these issues affect data integrity (except perhaps for LSI issue). Quote Link to comment
cal87 Posted April 29, 2012 Share Posted April 29, 2012 So a little more info on parity speeds. Like i posted earlier. 4.7 was ~55MB/sec 5.0RC1 was ~17.50MB/sec I've now dropped back to 5.0B14. Now parity check is running at ~50MB/sec. So RC1 certainly has some issues. I am getting the same 17.50MB/sec speed, but for me this was with B14 as well as RC1. I am running C2SEA with a couple of AOC-SASLP-MV8 cards in a RPC-4224 case. Was getting better speeds a few versions back. Quote Link to comment
speeding_ant Posted April 29, 2012 Share Posted April 29, 2012 Here is the list of issues gleened from this thread: -several users: lsi based controllers have issue accessing spun down disks -sacretagent: wants 24 disk support -kingping: reports "hdparm sending ioctl 2285 to a partition!" filling system log -crankbearing: wants to know what version of lm_sensors is being used [it's 3.1.2] -peter_sm: has issue where subtitles disappear using NFS -dlandon: clicking utils/system log constantly refreshes using IE9 -darkside40: samba 3.6.3 has a vulnerabilty which apparently has a patch somewhere -marky (and others): webGui freezing, especially during parity sync -burnaby_boy, _vern_: writes via network much slower vs. -beta14 -Fred: can't get flash share to export Here's my experience: The system log refreshes with Firefox & Safari as well. FTP works, but can't see files. Just did a parity rebuild, 17.5MB/s, did grow slowly to 25MB/s. AMD CPU & Gigabyte GA-870A-UD3 Cheers! Quote Link to comment
Influencer Posted April 29, 2012 Share Posted April 29, 2012 Limetech: The Samba issue has a patch released here: http://www.samba.org/samba/history/security.html The vulnerability is also fixed in the latest stable release, 3.6.4 . Not sure if you saw my previous post, the iotcl writing to syslog is harmless, and according to Linus will be removed completely in upcoming updates, it was used to debug another vulnerability that gave access to system memory. The rest of the issues are way above my paygrade, Quote Link to comment
Ice_Black Posted April 29, 2012 Share Posted April 29, 2012 How can you be sure that is kernel issues? I suspect Unraid dont work well the new kernal and Unraid need to be updated to be compatitable with new kernel? Quote Link to comment
sidezero Posted April 29, 2012 Share Posted April 29, 2012 I just ran into a bug as well. Found no network connectivity to my server so logged into the remote console and found the following spamming the console: e1000e: eth0 NIC Link is UP 1000 Mbps Full Duplex, Flow Control: Rx/Tx e1000e 0000:03:00.0: eth0: Reset adapter It's just spamming this about every second or so and has for a few hours now. Quote Link to comment
jumperalex Posted April 29, 2012 Share Posted April 29, 2012 Here is the list of issues gleened from this thread: -kingping: reports "hdparm sending ioctl 2285 to a partition!" filling system log same here ... but i must admit I'm also still running add-on: unmenu and simple features. Quote Link to comment
dlandon Posted April 29, 2012 Share Posted April 29, 2012 Here is the list of issues gleened from this thread: -kingping: reports "hdparm sending ioctl 2285 to a partition!" filling system log same here ... but i must admit I'm also still running add-on: unmenu and simple features. I just started a preclear and I seem to be getting these errors. I am also using simple features. Quote Link to comment
dlandon Posted April 29, 2012 Share Posted April 29, 2012 Here is the list of issues gleened from this thread: -kingping: reports "hdparm sending ioctl 2285 to a partition!" filling system log same here ... but i must admit I'm also still running add-on: unmenu and simple features. I just started a preclear and I seem to be getting these errors. I am also using simple features. CONFIRMED: I took out Simple Features and the log entries stopped. Quote Link to comment
speeding_ant Posted April 29, 2012 Share Posted April 29, 2012 Just to confirm what Influencer is saying, those log entries are purely a debug entry not an error Quote Link to comment
dlandon Posted April 29, 2012 Share Posted April 29, 2012 Here is the list of issues gleened from this thread: -kingping: reports "hdparm sending ioctl 2285 to a partition!" filling system log same here ... but i must admit I'm also still running add-on: unmenu and simple features. I just started a preclear and I seem to be getting these errors. I am also using simple features. CONFIRMED: I took out Simple Features and the log entries stopped. Tracked it down to email notifications in Simple Features. Stopping email notifications stops the log entries. Quote Link to comment
Recommended Posts
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.