Adding a second volume to Javelin - how to make it visible to network?

[HansSchulze]

I created a single volume 250GB RAID0 array, which gets the 8 standard exported directories (Download, Music, etc), internally called VOLUME1. I then added a 2TB WD20EARS as a new RAID0 array, it formatted (was NTFS, no warning before formatting), and shows up as VOLUME2. Quota and other menus show both volumes as ready and idle. How do I write to the second volume over the network? I don't see any WinXP visible exports that I can map drive letters to. What am I missing? Do I need to add some symbolic links to make the second volume visible?

I am not quite fond of the folder layout, I much prefer DLink's style where they just export all the volumes by VOLUME_1 etc and you create your own subfolders and shares. I also have had several severe raid issues in the past and would rather keep my files on backed up on 2 different servers, and on unrelated drives.

Note that the previous WD20EARS single drive somehow failed, will RMA, just seeks and isn't detected by PC SBIOS. This was VOLUME2. Then I plugged in the 250GB, worked on that for a while, and found that in some places the 250GB was called VOLUME2, and others it was called VOLUME1. When I replaced the WD20EARS, and deleted all volumes, and recreated them, I now have VOLUME1 and VOLUME2 names that make sense. Bug! :)

[HansSchulze]

I did the following, just want to know if this is what is expected to be necessary (what about those people who don't know Linux?):
- installed rooter.ppg as application and rebooted
- telnet javelin 2380
- mkdir /VOLUME2/VIDEO
- chmod 777 /VOLUME2/VIDEO (allows access to everyone to create files in the new folder)
- cd /VOLUME1/VIDEO
- rm * (delete all my previous files there)
- rmdir .RECYCLER (hidden directory)
- cd ..
- rmdir VIDEO
- ln -s /VOLUME2/VIDEO VIDEO (connects the second drive as the first one's VIDEO directory)
Note that these commands could have been done through the http://javelin:4080 MLDonkey web page as well. Also, another VIDEO subfolder could have been created instead of emptying that whole area on the first drive. But since my server is empty, I can experiment at will :)

[HansSchulze]

Note that this method no longer makes the VIDEO folder visible to the web GUI, so you can't change permissions there, you have to do it using chmod.

[HansSchulze]

It seems that the web Admin / FileNPrint / FileSharing / Create dialog doesn't recognize my VOLUME2, otherwise I could have skipping the above. Given the trouble of having the first VOLUME1 and VOLUME2 instances get messed up due to a disk failure, I am half certain that there are still some settings that haven't been reset about these volumes.

I did hear the Javelin start/stop one drive 3 times within a minute, but since I had the LED's off (late at night) I couldn't tell which one or why, so at least I turned on the LEDs permanently. I noticed that the file creation dates were off, so I turned off the daylight savings time, and the system time now matches my PC's. I also lost my telnet connection, so I decided to reboot. My log showed a number of errors (see below) so I ran an file system check from the web page. It seems that the check didn't report anything to the log for the first drive, yet checking the second drive caused the volume not to be properly remounted and now I show a red system LED. I rebooted again, and everything seems ok. We really have to get to the bottom of the errors, and improve the detail and size of the visible log window for the next version of software.

Nov 26 01:47:39 WARNING Task 30 timeout on disk 1 at LBA 0x0880 (Length 0x80)
Nov 26 01:46:34 WARNING Task 30 timeout on disk 1 at LBA 0x0580 (Length 0x80)
Nov 26 01:00:48 WARNING Task 30 timeout on disk 1 at LBA 0x09d00 (Length 0x80)
Nov 26 00:59:43 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 17:25:50 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 16:47:09 WARNING Task 30 timeout on disk 1 at LBA 0x0cf09800 (Length 0x80)
Nov 25 15:50:13 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 15:16:13 WARNING Task 30 timeout on disk 1 at LBA 0x0cf04480 (Length 0x80)
Nov 25 11:05:25 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 10:14:49 WARNING Task 30 timeout on disk 1 at LBA 0x0cf0f500 (Length 0x80)
Nov 25 08:30:07 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 07:38:06 WARNING Task 30 timeout on disk 1 at LBA 0x0580 (Length 0x80)
Nov 24 23:24:13 WARNING Task 30 timeout on disk 1 at LBA 0x0cf33d80 (Length 0x80)
Nov 24 23:06:52 WARNING Task 30 timeout on disk 1 at LBA 0x09d00 (Length 0x80)

I presume that disk 1 is the same disk1 that is listed in the web page, but that drive is empty except for a few OS folders (300 MB/250GB). All my data is on the second drive. I haven't found a way to cause it to do an xfs_check or fsck.xfs or badblocks (not included in the binaries) nor a surface test to see which drive corroborates the timeout errors. Then again, this second drive has less than a few hundred hours on it, it shouldn't be dying, essentially used twice for a drive-to-drive backup. Confused.

Now if only WD would actually correct the mfg date in their database per an email request I sent them so I can RMA the drive and finish my backup...:mad:

[kevinkar]

My log showed a number of errors (see below)

Nov 26 01:47:39 WARNING Task 30 timeout on disk 1 at LBA 0x0880 (Length 0x80)
Nov 26 01:46:34 WARNING Task 30 timeout on disk 1 at LBA 0x0580 (Length 0x80)
Nov 26 01:00:48 WARNING Task 30 timeout on disk 1 at LBA 0x09d00 (Length 0x80)
Nov 26 00:59:43 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 17:25:50 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 16:47:09 WARNING Task 30 timeout on disk 1 at LBA 0x0cf09800 (Length 0x80)
Nov 25 15:50:13 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 15:16:13 WARNING Task 30 timeout on disk 1 at LBA 0x0cf04480 (Length 0x80)
Nov 25 11:05:25 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 10:14:49 WARNING Task 30 timeout on disk 1 at LBA 0x0cf0f500 (Length 0x80)
Nov 25 08:30:07 WARNING Task 30 timeout on disk 1 at LBA 0x011b80 (Length 0x80)
Nov 25 07:38:06 WARNING Task 30 timeout on disk 1 at LBA 0x0580 (Length 0x80)
Nov 24 23:24:13 WARNING Task 30 timeout on disk 1 at LBA 0x0cf33d80 (Length 0x80)
Nov 24 23:06:52 WARNING Task 30 timeout on disk 1 at LBA 0x09d00 (Length 0x80)

I am not an expert but I think the drives you are using, 2TB WD20EARS, may be the problem. I just checked and they appear to be Western Digital Caviar *Green* drives which are never a good thing to use in any RAID even if the manufacturer of the drives or the RAID say they are OK. "Green" drives are prone to slow down or sleep to conserve electricity and that's not a good thing in a RAID environment where the controller is expecting the data to be available, read/write, at a moment's notice.

The errors you are seeing are very similar, if not the same as, the same errors a friend of mine saw when using FreeNAS and similar, if not the same, green drives. When he swapped out the drives for non-green drives, they responded better and the errors went away.

I have been using 3TB Seagate Barracuda drives and they are pretty speedy and I never see these errors.

Not that you want to go out and buy more drives but you may want to look into this green drive issue if you have some time to spare.

[BadIntentions]

Green drives cause timeouts by responding too slow. The Javelin has an enterprise grade raid controller (well, it claims to be one anyway) and it is impatient. It works best with drives that support Time Limited Error Recovery (TLER) or vendor equivalent.

[HansSchulze]

Appreciated the feedback. Unfortunately the last two RAIDs that failed were Baracudas, and I can give the remaining drives to you if you like them that much. They grew hard errors like mold on bread. And the Smart function wasn't at all helpful.
I am not even sure which drive is causing the error from the message, the 250GB first drive (WD black) or the 2TB Green second drive. The range of block numbers indicate <250GB.
I am not sure that the electronics of the Javelin isn't toasting the drives... as the first one is completely dead already, no longer detects on a PC, after being powered up a week in a PC, and a few days in the Javelin. Note that the toasted drive was solo in the case, maybe the power supply didn't like such a low load - I hope it's something the Javelin qual engineers have checked.
Two of the same model drives have been in a DLINK DNS-323 for 3/4 year without any issues. And I kinda like that it's 7W on idle.
The reason for the NAS is storage that gets used one or two hours a day, then sleeps the rest of the time. I am not using RAID, just JBOD off-line backed up storage.
If I wanted more speed, I would use a Promise controller in a PC box, and get 100 MB/sec all the time. I don't need more than 5 MB/sec for most usage, and only more speed for faster backups.
As far as the Green drives go, they should only spin down when the controller asks them to. Going into and out of "standby" shouldn't take more than a few ms. which is fine. I don't know that the spin-up time is much different than the Baracudas, about 5 seconds.

[kevinkar]

Hmmm, interesting. I have purchased nothing but Barracudas for probably the last 12 years and have not yet had one fail. There IS one I am a bit worried about but it's one of the oldest and is being replaced by an SSD (it's currently my main boot drive) so that's not a big deal.

So far, the 4 3TB Barracudas in my Javelin are working fine though they are JBOD and are essentially in a "read only" state at the moment (they serve DVD ISO, MP4, and MP3 files to any device on my network requesting them).

And you are right, green drives shouldn't really be "slow" but they just seem to tend to be and cause the timeout errors no matter how minute that time out may be.

My suggestion is to never go cheap on drives. Always buy the best drive possible. It's no fun trying to recover data. Copy to other locations and update the backups often!