CDLehner Posted March 23, 2011 Author Share Posted March 23, 2011 If you still can access the web interface. You'll have to shutdown another way. Do you have the powerdown addon installed? If not see this post: http://lime-technology.com/forum/index.php?topic=1905.msg13461#msg13461 OK, I don't mean to be rude...but for like the 4th time, I do NOT have access to the web GUI. If I did, there wouldn't really be an issue; or at least not this issue. I have unmenu; does that give me the powerdown addon? If not, then I don't have it. Is there really no way to shut this thing down, even with a keyboard and monitor at the server?? I've been so damn tempted to just hard powerdown, get back up and online. No one has suggested it's safe to do so, so I'll just ask the question; what would happen? CD Link to comment
Joe L. Posted March 23, 2011 Share Posted March 23, 2011 If you still can access the web interface. You'll have to shutdown another way. Do you have the powerdown addon installed? If not see this post: http://lime-technology.com/forum/index.php?topic=1905.msg13461#msg13461 OK, I don't mean to be rude...but for like the 4th time, I do NOT have access to the web GUI. If I did, there wouldn't really be an issue; or at least not this issue. I have unmenu; does that give me the powerdown addon? If not, then I don't have it. Is there really no way to shut this thing down, even with a keyboard and monitor at the server?? I've been so damn tempted to just hard powerdown, get back up and online. No one has suggested it's safe to do so, so I'll just ask the question; what would happen? CD If you have access to unMENU you can use the button on its "Array Management" link on it to stop the array. Once it is stopped, you can then just use the power switch to shut the server down. Link to comment
CDLehner Posted March 23, 2011 Author Share Posted March 23, 2011 If you still can access the web interface. You'll have to shutdown another way. Do you have the powerdown addon installed? If not see this post: http://lime-technology.com/forum/index.php?topic=1905.msg13461#msg13461 OK, I don't mean to be rude...but for like the 4th time, I do NOT have access to the web GUI. If I did, there wouldn't really be an issue; or at least not this issue. I have unmenu; does that give me the powerdown addon? If not, then I don't have it. Is there really no way to shut this thing down, even with a keyboard and monitor at the server?? I've been so damn tempted to just hard powerdown, get back up and online. No one has suggested it's safe to do so, so I'll just ask the question; what would happen? CD If you have access to unMENU you can use the button on its "Array Management" link on it to stop the array. Once it is stopped, you can then just use the power switch to shut the server down. No Joe, sorry...I didn't mean I had access to unMenu; I don't. Dgaschk asked me if I had the powerdown addon installed, and I was saying I have unMenu installed, and asked if that might include this addon. CD Link to comment
Joe L. Posted March 23, 2011 Share Posted March 23, 2011 OK, I now understand better. Ys, the clean powerdown is installable via unMENU's package manager, but only you will know if you installed it. In any case, a quick press of the power button would trigger it if it was. If not, and if "Control-Alt-Delete" does not work from the system console the only thing you can do is power down (by the switch on the power supply, or by pressing and holding down the power button on the server case, or by un-plugging the array from the wall socket) Link to comment
SSD Posted March 23, 2011 Share Posted March 23, 2011 Run the sync command before shutting down your server. At least that will force unwritten blocks back to the disk. Link to comment
CDLehner Posted March 23, 2011 Author Share Posted March 23, 2011 OK, I now understand better. Ys, the clean powerdown is installable via unMENU's package manager, but only you will know if you installed it. In any case, a quick press of the power button would trigger it if it was. If not, and if "Control-Alt-Delete" does not work from the system console the only thing you can do is power down (by the switch on the power supply, or by pressing and holding down the power button on the server case, or by un-plugging the array from the wall socket) Thanks Joe. I'm at work now, so I'll be trying this tonight...so let me try and get my ducks in a row. You're saying the powerdown addon is part of the unMenu package, but it would have needed to be installed. I don't think I have, but you're saying if I do a quick power button press (as opposed to press and hold), it'll shut-down the server if it is installed. If not, no harm no foul? As for control-alt-delete; obviously I know how it works in a Windows environment. What exactly am I looking for if I do it at the unraid server? If neither of that works, you're saying go ahead and do a hard powerdown? What are the risks? CD Link to comment
CDLehner Posted March 23, 2011 Author Share Posted March 23, 2011 Run the sync command before shutting down your server. At least that will force unwritten blocks back to the disk. Sorry; can I get the command string? I'm at login now. CD Link to comment
SSD Posted March 23, 2011 Share Posted March 23, 2011 Very simple: sync It can take a minute or so to complete depending in if it has to spin up disks and how much data is buffered. I'd also recommend ctrl-alt-del to big-red-switch if it works. Just make sure sync has finished. Link to comment
Joe L. Posted March 23, 2011 Share Posted March 23, 2011 Run the sync command before shutting down your server. At least that will force unwritten blocks back to the disk. Sorry; can I get the command string? I'm at login now. CD If you are at the login prompt and can log in as "root" you have lots of possibilities available to you. If you can get to a linux command prompt you can try typing this series of commands /root/samba stop umount /mnt/disk1 umount /mnt/disk2 umount /mnt/disk3 as needed based on how many data disks you have installed (note: it is "umount" not "unmount") umount /mnt/cache If you can un-mount all the disks, you can then stop the array by typing: /root/mdcmd stop followed by reboot or poweroff If a disk cannot be un-mounted, it will be because it is "busy" A disk will be busy if a file on it is open or if it is the current directory of process. You can see the process IDs of the processes holding disks busy by typing: /usr/bin/fuser -mv /mnt/disk* /mnt/user/* /mnt/cache you can then kill any given process by typing: kill NNNNN where NNNNN = the process ID. Then after killing any processes holding any disk busy, un-mount them, then stop the array once all the disks are un-mounted with the command listed above. Joe L. Link to comment
CDLehner Posted March 23, 2011 Author Share Posted March 23, 2011 Yikes! Lots to try I'll have my work cut out for me later. I'd really love to try and get back online tonight. It's been 2 days now...simply because I tried to preclear a disk! CD Link to comment
CDLehner Posted March 23, 2011 Author Share Posted March 23, 2011 Run the sync command before shutting down your server. At least that will force unwritten blocks back to the disk. Sorry; can I get the command string? I'm at login now. CD If you are at the login prompt and can log in as "root" you have lots of possibilities available to you. If you can get to a linux command prompt you can try typing this series of commands /root/samba stop umount /mnt/disk1 umount /mnt/disk2 umount /mnt/disk3 as needed based on how many data disks you have installed (note: it is "umount" not "unmount") umount /mnt/cache If you can un-mount all the disks, you can then stop the array by typing: /root/mdcmd stop followed by reboot or poweroff If a disk cannot be un-mounted, it will be because it is "busy" A disk will be busy if a file on it is open or if it is the current directory of process. You can see the process IDs of the processes holding disks busy by typing: /usr/bin/fuser -mv /mnt/disk* /mnt/user/* /mnt/cache you can then kill any given process by typing: kill NNNNN where NNNNN = the process ID. Then after killing any processes holding any disk busy, un-mount them, then stop the array once all the disks are un-mounted with the command listed above. Joe L. OK, I feel like I'm making progress here. I was able to login to root, and get to a linux command. I ran /root/samba stop and I was able to umount all the disks (I could see their activity, so I know it was working). I'm very anxious to do the reboot or poweroff, and see if I can get back online again. A couple last questions, before it's too late. One, do I need to do an umount for the parity (umount /mnt/parity)? Two, do I still need to do the sync suggested above? Thanks, CD Link to comment
SSD Posted March 23, 2011 Share Posted March 23, 2011 No umounting of parity. sync won't hurt, but don't think it will be necessary. You do need to do the /root/mdcmd stop command. If that ends normally, you should be golden to reboot. Link to comment
CDLehner Posted March 23, 2011 Author Share Posted March 23, 2011 No umounting of parity. sync won't hurt, but don't think it will be necessary. You do need to do the /root/mdcmd stop command. If that ends normally, you should be golden to reboot. OK, didn't worry about unmounting the parity...did run the /root/mdcmd stop, and did a powerdown. Powered back up and everything appears to be "normal" again. I was going to ask if you guys still needed/wanted to see a syslog or anything, but it doesn't appear there's anything to look at. What, it got overwritten with the fresh boot? CD Link to comment
SSD Posted March 24, 2011 Share Posted March 24, 2011 Did you do the cp command from THIS post earlier in the thread: cp /var/log/syslog /boot/syslog.txt If so, the syslog is on the flash share in the file syslog.txt. Zip it and post it. Link to comment
CDLehner Posted March 24, 2011 Author Share Posted March 24, 2011 Did you do the cp command from THIS post earlier in the thread: cp /var/log/syslog /boot/syslog.txt If so, the syslog is on the flash share in the file syslog.txt. Zip it and post it. I did...but that's the thing; there doesn't appear to be that much to it. Certainly not the lines and lines that I'm used to in a syslog. But here it is; didn't even need to zip it...lol. Tell anyone anything? CD syslog.txt Link to comment
dgaschk Posted March 24, 2011 Share Posted March 24, 2011 IT shows a problem with sdg and the Ethernet going down then up. Post a new syslog. Link to comment
SSD Posted March 24, 2011 Share Posted March 24, 2011 How'd this wind up in the trash bin? I can only guess that your computer ultimately killed the logger itself, and the somehow it was restarted and you got a few lines in the log afterwards. I am no expert and reading these logs, but it appears your computer was having a problem with disk ATA6 (you can look at your current syslog - the ATA numbers would likely be the same, but no guarantees). My guess would be, however, that based on the fact you were preclearing a disk, that ATA6 was the disk you were preclearing. Messages like "hard resetting link" and "SATA link down" indicate that the computer has lost contact with the drive and is attempting "CPR" to try to get it talking again. But it is not working. Most common reasons for losing contact is a bad or lose SATA or power cable. But there have been some cases of drives, esp. the EARS drives that have had their jumpers removed, stopping responding and requiring a power cycle to start talking again. I am not he expert on these things. Suggest we get Joe L. and/or Peter (lionelhutz) to take and look and give their thoughts. But perhaps you can comment on what type of drive you were preclearing and its "history". You also should check all of the power and data connections to the drive. Link to comment
CDLehner Posted March 24, 2011 Author Share Posted March 24, 2011 How'd this wind up in the trash bin? I can only guess that your computer ultimately killed the logger itself, and the somehow it was restarted and you got a few lines in the log afterwards. I am no expert and reading these logs, but it appears your computer was having a problem with disk ATA6 (you can look at your current syslog - the ATA numbers would likely be the same, but no guarantees). My guess would be, however, that based on the fact you were preclearing a disk, that ATA6 was the disk you were preclearing. Messages like "hard resetting link" and "SATA link down" indicate that the computer has lost contact with the drive and is attempting "CPR" to try to get it talking again. But it is not working. Most common reasons for losing contact is a bad or lose SATA or power cable. But there have been some cases of drives, esp. the EARS drives that have had their jumpers removed, stopping responding and requiring a power cycle to start talking again. I am not he expert on these things. Suggest we get Joe L. and/or Peter (lionelhutz) to take and look and give their thoughts. But perhaps you can comment on what type of drive you were preclearing and its "history". You also should check all of the power and data connections to the drive. sdg/ata6/disk6 was the disk I was trying to preclear. It's an old, pos Seagate 500G. Truth be told, I thought it was bad...but didn't see the harm in putting it through a preclear, and seeing if I got lucky. Guess I was wrong; guess there was harm in running the preclear, on a disk I suspected of being bad. CD Link to comment
SSD Posted March 24, 2011 Share Posted March 24, 2011 sdg/ata6/disk6 was the disk I was trying to preclear. It's an old, pos Seagate 500G. Truth be told, I thought it was bad...but didn't see the harm in putting it through a preclear, and seeing if I got lucky. Guess I was wrong; guess there was harm in running the preclear, on a disk I suspected of being bad. CD The drive may very well be bad. But nothing here in the log to point the finger at it yet. It is not typical that a bad drive will just stop responding, but a bad drive can do anything, so we don't know for sure. Can you run a SMART report on it? smartctl -d ata -a /dev/sdg Link to comment
dgaschk Posted March 24, 2011 Share Posted March 24, 2011 A bad drive can have unexpected consequences. Any unusual things can happen. Usually removing the bad drive fixes things. That is unless corrupted data is written to remaining storage devices. Link to comment
CDLehner Posted March 24, 2011 Author Share Posted March 24, 2011 sdg/ata6/disk6 was the disk I was trying to preclear. It's an old, pos Seagate 500G. Truth be told, I thought it was bad...but didn't see the harm in putting it through a preclear, and seeing if I got lucky. Guess I was wrong; guess there was harm in running the preclear, on a disk I suspected of being bad. CD The drive may very well be bad. But nothing here in the log to point the finger at it yet. It is not typical that a bad drive will just stop responding, but a bad drive can do anything, so we don't know for sure. Can you run a SMART report on it? smartctl -d ata -a /dev/sdg I'll try. CD Link to comment
CDLehner Posted March 24, 2011 Author Share Posted March 24, 2011 A bad drive can have unexpected consequences. Any unusual things can happen. Usually removing the bad drive fixes things. That is unless corrupted data is written to remaining storage devices. I want to make it clear...as I was running preclear...this wasn't in the array yet. So you're saying even if I put a bad drive in the server, just for the purposes of seeing if it's bad, can corrupt the array? CD Link to comment
Joe L. Posted March 24, 2011 Share Posted March 24, 2011 A bad drive can have unexpected consequences. Any unusual things can happen. Usually removing the bad drive fixes things. That is unless corrupted data is written to remaining storage devices. I want to make it clear...as I was running preclear...this wasn't in the array yet. So you're saying even if I put a bad drive in the server, just for the purposes of seeing if it's bad, can corrupt the array? CD Not very likely... If it locks up the SATA controller it could stop the array but I've never seen a disk failure affect data on another. The errors in the system log from a failing drive could fill all available RAM and eventually cause the kernel out-of-memory process to terminate idle processes in an attempt to free memory. I think that might have occurred in your server. One process that is idle is "emhttp" (the process that supplies the web-management interface) Once it is killed you'll seemingly lose control of the server. Joe L. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.