gfjardim

###2019.04.22 - Fix: remove ncurses dependency; - Fix: save resume information to flash drive every 15 minutes to avoid NAND memory wear.

Well, my flash drive block is 4k sized. So 15 minutes between writes will result in 160 writes over 40 hour span, which result in 640KB written to the flash drive. I think this is reasonable, isn't it?

@s.Oliver, I couldn't decide it for myself, so I'll open this for debate. Guys, which is the desirable time lapse to write the resume info to the flash drive?

Yes, the Preclear plugin writes a resume file to the USB stick about three times a minute, and that file has about 400 bytes. If you have a 40 hour Preclear session, you will have a total of 2.74MB written to your flash drive at the end of it. If your flash drive has 32GB and each cell can be written 10.000 times, you can easily write Terabytes to it until it fails.

I need to see the preclear plugin log (see main OP.)

Download all these files into your flash drive (/boot/config/plugins/preclear.disk/) and reboot the server. https://github.com/gfjardim/unRAID-plugins/raw/master/source/packages/ncurses-6.1_20180324-x86_64-1.txz https://github.com/gfjardim/unRAID-plugins/raw/master/source/packages/libevent-2.1.8-x86_64-1.txz https://github.com/gfjardim/unRAID-plugins/raw/master/source/packages/tmux-2.7-x86_64-1.txz https://github.com/gfjardim/unRAID-plugins/raw/master/source/packages/utempter-1.1.6-x86_64-2.txz https://github.com/gfjardim/unRAID-plugins/raw/master/archive/preclear.disk-2019.02.12c.md5 https://github.com/gfjardim/unRAID-plugins/raw/master/archive/preclear.disk-2019.02.12c.txz https://github.com/gfjardim/unRAID-plugins/raw/f3d09609407cf559666e8e9b1aa17ba9d5324ca3/plugins/preclear.disk.plg

Which Unraid version are you using?

And you can update the plugin even if you already have preclear sessions running.

Thinking about your request, I found a way to easily implement this, so it's added in version 2019.02.12. Just go to to Tools > Preclear Disk and, in the title bar, click on the tachometer icon. Ps: this will not survive a reboot.

It's a small bug I just patched. Thanks for reporting! To mitigate this you can set the script to notify you once it finishes, then you can start another batch of preclear sessions.

You can try to guest the SMART device type using a function like get_smart_type in this file.

This way: <FILE Name="/boot/packages/&gittools;" Run="upgradepkg --install-new" Max="6.6.6"> <URL>http://mirrors.slackware.com/slackware/slackware64-14.2/patches/packages/&gittools;</URL> <MD5>&gittoolsMD5;</MD5> </FILE> And the plugin should download the package elsewhere. If you remove the plugin and it leaves the package as a leftover, Unraid will continue to install it upon boot.

@olehj, is slot assignment persistent to disk slot change? If remove disk A from slot 1 and disk b from slot 2, then move disk A to slot 2 and add disk C to slot 1, do disk A appear on slot 2 and disk C on slot 2?

Strange that it has no surface errors, so could be bad SATA cable, bad cache memory or bad disk controller.

@John_M, can you try installing the latest version of the plugin?

@mrvilla, please send me the preclear plugin log (see the OP).

I like to run 2 cycles, but that's me. I saw nothing that indicates you got bad drives. Erase writes random data, and it's meant to erase disks prior to sell/retirement. In your case, Clear will do just fine.

For disk ST12000NM0007-2A1101_ZJV1HWVE and ST12000NM0007-2A1101_ZJV1GVM5 you have toggled the Enable Testing (just for debugging) switch, which only appear if Help is toggled. This option only read a portion of the disk and it's meant to allow me to proper test some features without having to wait a full prelear session. You have to run preclear on them again, without the testing option. The disk ST3250310AS_6RY9FV4D had a successful preclear session. The other two disks preclearing now are correctly invoked, you just have to sit and wait.

If you send me your preclear log (see the OP) I can tell you if it was successful or not.

Yep. I changed the logic on the script too, so it can write the whole resume information to the flash drive and then rename it.

I think your server locked while the script was writing the resume information on the boot flash drive. You must cancel the resume popup and start a new instance of preclear to that drive.

Please send me the plugin log (see the first post of this topic).

And if he's having problems with NETBIOS names, he can always use IP addresses instead.

The problem with badblocks is that it actually doesn't detect bad blocks. This program was relevant when media errors were tracked by the filesystem. Nowadays, bad blocks are usually retired by S.M.A.R.T. routines and their position remapped by drive's firmware. If a sector is unreadable, either dd or badblocks will be aborted. The trick is to write data, read it and, as you said, compare some critical S.M.A.R.T attributes from the beginning with those same atributes at the end.

Instead du you could use docker save for images and docker export for containers. Will work even on those cases du doesn't work.