April 2, 201115 yr Losing network connection. Unable to access via IP (putty/telnet). 2nd occurrance, had to hard boot before. Looking at the switch, both lights were flickering very rapidly. I have unplugged it from the switch and will wait a little bit to see if the system console will respond. like it did for smakovits here http://lime-technology.com/forum/index.php?topic=11826.msg112685#msg112685 Was preclearing a drive, last e-mail showed 50% of pre-read at 5:15am. Hooked up monitor and keyboard to server (which I had tail -f /var/log/syslog running). Realize my mistake now, should have had it in a telnet window instead of on the server so I could at least copy/paste. Did see an unable to allocate memory error before the monitor went black...not getting the "screen" to appear again. SLUB I think. Checking to see if anyone has any other ideas before I just hard boot the machine, then try it again with a telnet session running tail -f on the syslog.
April 2, 201115 yr Author So I found this regarding log explosion http://lime-technology.com/forum/index.php?topic=3352.msg114003#msg114003 /etc/rc.d/rc.syslog stop mkdir /tmp/log mount tmpfs /tmp/log -t tmpfs cp /var/log/syslog /tmp/log/ rm /var/log/syslog ln -s /tmp/log/syslog /var/log/syslog /etc/rc.d/rc.syslog start I guess this would go at the end of the go script? Could (or should) I change /tmp/log to a location on the array? That way in case of hard boot, the syslog is not lost? Or would there be a better way to have the syslog go to a location on the array? I know the syslog gets copied during gracefull shutdowns, but wanting to find a way in case of hard boots/crash of not losing the syslog for troubleshooting.
April 2, 201115 yr Author ok...rebooted...got telnet window (on windows machine) doing tail -f on the syslog. restarted the preclear_disk.sh -A -m email@address -M 4 /dev/sdg hasn't locked up but am seeing some red errors in the syslog SLUB: Unable to allocate memory on node -1 (gfp=0x20) syslog.txt
April 2, 201115 yr Author ok...seems good timing that I posted the syslog in the message just above. lost connection to IP address and no response with the system monitor/keyboard The additional lines to the syslog from that posted above are: Apr 2 12:37:45 Tower kernel: 5030 pages reserved Apr 2 12:37:45 Tower kernel: 243596 pages shared Apr 2 12:37:45 Tower kernel: 217066 pages non-shared Apr 2 12:59:23 Tower kernel: __ratelimit: 14 callbacks suppressed Apr 2 12:59:23 Tower kernel: swapper: page allocation failure. order:0, mode:0x4020 Apr 2 12:59:23 Tower kernel: Pid: 0, comm: swapper Not tainted 2.6.32.9-unRAID #8 Apr 2 12:59:23 Tower kernel: Call Trace: Apr 2 12:59:23 Tower kernel: [<c104d06f>] __alloc_pages_nodemask+0x3fb/0x42f Apr 2 12:59:23 Tower kernel: [<c1256939>] ? tcp_send_delayed_ack+0xb3/0xb8 Apr 2 12:59:23 Tower kernel: [<c106841c>] __slab_alloc+0x13e/0x425 Apr 2 12:59:23 Tower kernel: [<c1068d8f>] __kmalloc_track_caller+0x86/0xcf Apr 2 12:59:23 Tower kernel: [<c1229a38>] ? __netdev_alloc_skb+0x17/0x34
April 2, 201115 yr This looks like a memory issue. Either your running out of memory due to an add-on or there is some bad RAM. Run a memtest overnight.
April 2, 201115 yr Author This looks like a memory issue. Either your running out of memory due to an add-on or there is some bad RAM. Run a memtest overnight. stupid question. can you run memtest while the array is up? or does it need to be offline?
April 2, 201115 yr This looks like a memory issue. Either your running out of memory due to an add-on or there is some bad RAM. Run a memtest overnight. stupid question. can you run memtest while the array is up? or does it need to be offline? No, you cannot run it with the array online.
April 3, 201115 yr Author MEMTEST ran without errors (7 passes) Any thoughts on what next to check? What command to check how much memory is used/free, and what is using the memory? Also, how much memory does preclear_disk.sh usually require for a 2TB drive?
April 3, 201115 yr in my limited experience setting up my server, hard server crashes are a result of running out of memory... this could be due to syslog explosion (syslog grows too large and hogs all the ramdisk) or something else that's chewing up the memory. Preclearing a drive shouldn't req. too much memory, do you have 4gig ? that's what I thought I read. I have 4gig and was able to preclear 5 drives at a time. To check available memory, you use the "free" command. Just type in in a telnet window. Other option is to run "top", it lists all the processes that's running and it updates itself. Keep this running in a telnet window until it crashes and see what it says. You can also have in another telnet window the following command: "tail -f -n 20 /var/log/syslog" it will show the contents of syslog and update it when new entries are added. When the unRaid server crashes, you can see what the last few entries are.
April 4, 201115 yr Author 2G RAM top - 08:23:42 up 18:54, 4 users, load average: 5.10, 5.03, 5.63 Tasks: 105 total, 1 running, 104 sleeping, 0 stopped, 0 zombie Cpu(s): 5.8%us, 11.9%sy, 0.0%ni, 25.2%id, 48.8%wa, 3.5%hi, 4.8%si, 0.0%st Mem: 1813480k total, 1771308k used, 42172k free, 104672k buffers Swap: 0k total, 0k used, 0k free, 1307832k cached Is cached normally this high of a percentage?
April 4, 201115 yr 2G RAM top - 08:23:42 up 18:54, 4 users, load average: 5.10, 5.03, 5.63 Tasks: 105 total, 1 running, 104 sleeping, 0 stopped, 0 zombie Cpu(s): 5.8%us, 11.9%sy, 0.0%ni, 25.2%id, 48.8%wa, 3.5%hi, 4.8%si, 0.0%st Mem: 1813480k total, 1771308k used, 42172k free, 104672k buffers Swap: 0k total, 0k used, 0k free, 1307832k cached Is cached normally this high of a percentage? yes. Linux will use all available memory for disk buffer cache.
April 4, 201115 yr Author I've started preclear_disk.sh again (although using version 1.9 instead of .9.9 which I had prior) Have the following telnet sessions running from windows machine: 1. tail -f -n 100 syslog 2. top 3. free script that does date and free commands every minute to a log file. Hopefully I won't lock up, but if it does, hope all this info will help. Anything else that I might want to log?
April 4, 201115 yr Author UnRAID 4.7 CoolMaster-Centurion 590 BioStar A760G M2+, AMD Athlon II X2 245w 1 x 2GB DDR2 800 RAM CM GX-750W PSU 1x SuperMicro 5x3 CSE-M35T1 SanDisk Cruzer 4GB (Boot) HDD parity Hitachi_HDS72202_ disk1 ST32000542AS_ disk2 WDC_WD20EADS-00R_WD- disk3 WDC_WD20EADS-00R_WD- disk4 SAMSUNG_HD204UI_ disk5 ST32000542AS_ disk6 WDC_WD20EARS-00M_WD- disk7 WDC_WD20EARS-00M_WD- Preclearing a SAMSUNG_HD204UI_
April 4, 201115 yr It looks like you are running single 2GB stick on a stock Biostar A760G with the default BIOS settings. This board does not have a sideport memory and it will some (I believe 128MB in the default settings) from your existing RAM. I do not have this board and cannot provide you more info but there is a setting in the BIOS (if you cannot find it you may have an older BIOS) that will allow you to at least trim the amount taken for the video. And no harm to disable any unused functions (ser. and par. ports, audio, floppy drive, IDE controller too if yo do not plan to use any of the older PATA drives) And if you do plan to use extra dd-ons why not add another 2GB. And Joe L. can give us an idea for the memory footprint every preclear session requires (it has gone thru many revisions in the last half a year or so...)
Archived
This topic is now archived and is closed to further replies.