DSM-G600, DNS-3xx and NSA-220 Hack Forum

flashman1207 wrote:

Fonz...... If I go back to using the default syslogd.sh (thereby logging to /var/log/messages) won't this cause the disks to spin up? Won't it write to disk and not RAM?

/var/log is in RAM.

flashman1207 wrote:

Code:

EXT2-fs warning: maximal mount count reached, running e2fsck is recommended

If this message is about the RAM-disk, that's up to Dlink to fix. A fsck run should fix it for the other disks. After fsck, there should be no errors remaining (that's why you run it, right?). If fsck detects problems, it'll ask you whether it should fix them. Obviously, you should say 'y'.

flashman1207 wrote:

I tried the "mdadm -A /dev/md0 /dev/sd[ab]2" command but it said "/dev/md0 already active - cannot assemble it".

What are you doing? Are you trying to run the commands from my fsck page without booting the ramdisk? If you don't know what you're doing, stop now, power off and reboot. You might delete all your data.

flashman1207 wrote:

Thanks for all the replies, I feel I am learning a lot.

silversurfer..... would the format be

Code:

 syslogd_flags="-O /tmp/syslog.txt -l 5"

Will Log level 5 show the disk spin/down info?

Yes, that looks good. According to the info from fonz a log level of 4 should already be enough to see warnings. HD spin up and down messages come at warning level. So a level of 4 to 7 should catch HD messages.


Edith says:

BTW the message "EXT2-fs warning: maximal mount count reached, running e2fsck is recommended" only says that the file system was mounted for a certain number of times without being checked. That doesn't mean that there are errors in the file system and your data is going to be lost.
Usually normal Linux distributions will run a file system check at startup when the maximal mount count has been reached and after that they mount the file system. I didn't check if the NAS does this for any of the file systems it uses. Of course it is advised to check the file system to be sure that everything is ok.

Last edited by silversurfer (2008-07-27 19:46:32)

Thanks Silversurfer, will set it to 4....

Now as to this e2fsck thing, I'm all a dither about this. Can I run it without unmounting the disk? Saw somewhere that you needed to run funplug on USB to unmount. Or am I confused with something else. What exactly is the e2fsck command I should be running anyway? Also, I saw in fonz fsck page that running RAID may make a difference, am RAID and JBOD on same box.....

Thanks as always
Tony

bq041 wrote:

e2fsck is a command that checks ext2 file systems.  (e)xt(2) (f)ile (s)ystem (c)hec(k)  No, you cannot run it with the drives mounted.  While they are mounted, they are "in use" and the program cannot properly access the filesystem.  To do this, you need to boot the ramdisk and then run the program on the disks.  If you are using RAID, then you must first assemble the RAID array, then run the program on the array.

b1041,
   Thanks for the replies but things like 'boot the ramdisk' and 'assemble the RAID array' don't mean anything to me (in this context anyway). I understand that obviously I am out of my depth here but have a problem that needs resolving. Is there a step by step guide anywhere to doing this complete with exact commands for us newbies? I'm not completely non-techie (being an old mainframe sysprog)  but have only just started tinkering with Linux... that alone will have a few people chuckling I'm sure.......

Would installing the funplug on USB make my life somewhat easier?

Tony

bq041 wrote:

http://inreto.de/dns323/fsck/

Keep in mind bob is the name of fonz's DNS; substitute it with the name or ip address of yours.

Looking at the instructions am I correct in saying that because I do not have 'bash' (or even know what it is) that the command to reload would be:

== Booting ==

  telnet bob
  cd /ffp/fsck                          (or /mnt/HD_a2/fsck)
  ./no-bash-reload.sh

Thanks
Tony

Last edited by flashman1207 (2008-07-28 10:19:03)

silversurfer wrote:

flashman1207 wrote:

Would installing the funplug on USB make my life somewhat easier?
Tony

Installing ffp on a USB disk will definitely make your life easier once you've done it right. By "right" I mean running everything from USB disk or at least RAM disk. That way you can unmount the hard disks easily every time you want to without rebooting the system. A few services that access the hard disks have to be stopped before unmounting, on my system it's only the Samba daemon "smbd". To find out which service is accessing the hard disks use "lsof | grep HD"

There are several scripts in this forum to install ffp on USB but I would say that most of them need a few modifications to be ready for your system setup, so this is not a trivial task. It will probably be easier to start your NAS in "file system check mode" as described in http://inreto.de/dns323/fsck/.

Silversurfer,
   I'm going to look at the 'file system check mode' thing but before I start I don't really want to look at also installing the bash thing so, as per my previous update is ./no-bash-reload.sh the thing to use to reboot? Fonz says:

For reload.sh, you also need the bash shell. If you cannot install
bash, you can try no-bash-reload.sh instead of reload.sh.

it says 'try no-bash-reload.sh instead of reload.sh'. Anyone tried it?

Just on another note, after all this too-ing and fro-ing had a look at my /var/log/messages file and though it does show the disks spinning up and down (and the times) it doesn't indicate what is causing it. Most of the time it is around every 90 minutes or so (but not always exactly). I've attached the output from the lsof | grep command. Is there anything here that needs attention?

Thanks
Tony

---

Attachments:
output, Size: 5,291 bytes, Downloads: 359

@flashman
I had the same issue 2 months ago and like to give you some info to think about. Because some people at this forum reported problems like data loss and syncing new drives with their RAID1 array, I decided to change my configuration from RAID 1 with JBOD to 2 seperate disks, so I would not have to bother with this software RAID solution and all the knowledge needed to maintain it. To increase my data redundancy I run a backup program (rsnapshot) on my NAS every night, which will copy my data from volume 1 to volume 2 and keeps a history as well. I had the same problem with the fs check, so I decided to move the fun_plug to usb flash drive. After that I was able to stop certain processes and umount the hard disks and run the checks. For the USB drive, I umount the device as read-only and run the file system check. Make sure that you install cleanboot as well, this will prevent any possible future fs errors in case 1 of the disks can not be properly unmounted because some installed process keeps a file open. The USB drive can also help you to keep the harddisks spun down if you install additional programs in the future.

@silversurfer
I read that you write syslog.txt to usb. My logfile contains more then 1000 entries after a few days (every 10 minutes I receive an entry like: Jul 28 20:10:01 MEDIASERVER cron.notice crond[1576]: USER root pid 23300 cmd /usr/sbin/getdhcp&). I think this big amount of write cycles will wear out the USB flash memory within a few months. That is the reason why I write syslog.txt to ram. BTW, I didn't know there was a default size for the logging

Last edited by EnricoM (2008-07-28 21:57:00)

flashman1207 wrote:

But what services could be accessing the disks if I'm not actually using the box?

One way to find out is to kill process by process until it stops. I'd start with the fan controller.

flashman1207 wrote:

Is there anything I shouldn't kill ????

telnetd

The command fuser will give you a list of processes that access your drive(s).

fuser -m /mnt/HD_a2

With the option fuser -k, you can kill them all in one go... (beware not to kill telnetd, sshd and sh)

Last edited by EnricoM (2008-07-29 15:11:31)

Along the lines of being a newbie and drive spin down...

I have transmission running, but at the current time, nothing is being downloaded or seeding.  When I run 'fuster -m /mnt/HD_a2'  I get several PIDs back, 3 of which are for transmission-daemon.

If there is nothing active in transmission, should the PIDs appear?  If it is correctly appearing, why should transmission access the HD when there isn't downloading/seeding?

Just trying to learn..

Thanks