/greyhole/tasks' is marked as crashed and should be repaired

[ralfb]

Hi

I upgraded to the latest version of greyhole 2 days ago.

I also added a new 2 TB drive.

I tried to balance across the drives and received the error message above.

My O/S drive is now full with only 5% storage left, that's 3.5GB.

I think something is broken.

Would appreciate your help fixing.

Many thanks,

Ralf

here is the error message:

[ralf@padme ~]$ su
Password:
[root@padme ralf]# service hdactl restart
Stopping hdactl: [ OK ]
Starting hdactl: [ OK ]
[root@padme ralf]#
[root@padme ralf]# greyhole --balance
Can't describe tasks with query: DESCRIBE tasks - Error: Table './greyhole/tasks' is marked as crashed and should be repaired[root@padme ralf]#
[root@padme ralf]# greyhole --view-queue
Can't describe tasks with query: DESCRIBE tasks - Error: Table './greyhole/tasks' is marked as crashed and should be repaired[root@padme ralf]#

Also tried the following a few hours ago, still waiting:

mysql -u root -phda -e "repair table tasks" greyhole
mysql -u root -phda -e "repair table tasks_completed" greyhole
mysql -u root -phda -e "repair table settings" greyhole
greyhole --view-queue

root@padme ralf]# service hdactl restart
Stopping hdactl: [ OK ]
Starting hdactl: [ OK ]
[root@padme ralf]# mysql -u root -phda -e "repair table tasks_completed" greyhole

Also can not log into the dashboard, all is see is a Ruby on Rails error screen.

Error message:
Mysql::Error: Got error 28 from storage engine: SHOW FIELDS FROM `settings` (ActiveRecord::StatementInvalid)

[gboudreau]

The repair table commands you launched should fix your Greyhole problems.
Also to note: those commands are run automatically by the Greyhole service, when it starts.

Byt you have bigger problems, since other databases seems broken.
I'd ask in the Amahi IRC channel, or another post in the appropriate forum.

[ralfb]

Many thanks,

What happened was is that I copied a bunch of files to my HDA and then the OS drive filled up and greyhole stopped responding.

I updated to your latest version a day or so earlier.

Is there anyway to downgrade again? Would that help?

Any idea how other databases can break?

Best,

Ralf

[gboudreau]

Downgrading won't help. The problem isn't Greyhole, it's that you filled your system drive, and MySQL didn't like that. If it crashed, the tables it was working on when it crashed are now broken, and need to be repaired.

[ralfb]

Hi, it seems the system drive always fills up when I copy files to the HDA and then it shrinks again when the data is written to the greyhole pool.

At least this is my understanding based on my observation.

It seems that in this case the data was not moved from the system drive to the storage pool.

I don't know what happened.

Also when running

Code: Select all

mysql -u root -phda -e "repair table tasks_completed" greyhole

in terminal, the task never completes contrary to:

Code: Select all

mysql -u root -phda -e "repair table tasks" greyhole mysql -u root -phda -e "repair table settings" greyhole

both of which complete with a new prompt.

Does this shed any light on the problem?

The only significant things I did on the server are:
1) add one more 2 TB drive as per wiki http://wiki.amahi.org/index.php/Adding_ ... o_your_HDA
2) change the reserved space on storage drives to 2% from 5% as per http://bugs.amahi.org/issues/show/693 (left system drive UNTOUCHED.
3) upgrade greyhole as per http://forums.amahi.org/viewtopic.php?f=39&t=2325


many thanks
Ralf

[gboudreau]

The tasks_completed table is much larger than the other two. Running a repair of that table will take quite a while.
Leave the repair running until it completes.

On Amahi, the / partition is used as the Greyhole landing zone by default.
You can change that if you'd like, in order to not use the / partition, even temporarily.
A search for "change landing zone" in this Greyhole forum should give you the info you need.

[ralfb]

ok great, I'll leave mysql -u root -phda -e "repair table tasks_completed" greyhole running overnight now.

Let's see what happens.


After searching changing the landing partition, it seems I need access to the HDA dashboard to change the landing zone, which I can't do now as I can not access the HDA dashboard at the moment.

Is there a way to flush the current landing zone?

Thanks for the tip.

Ralf

PS.

I read some more of your support threads here.

Would it be simpler to delete amahi and reinstall it and then reconnect / remount the storage drives and create the shares again? (if yes, in which order?)

Would you happen to know if it is possible delete Amahi whilst keeping the Fedora install intact as I've installed VirtualBox with Windows on it and that was quite a challenge.

many thanks for your kind and patient support. It must get frustrating for you to find some of us are not at your level of understanding and must therefore occur as painfully slow to you to coach. I can get that. Anyway, thank you, and thank you.

Your Greyhole is an awesome invention!!! and why I chose Amahi over other NAS solutions.

Ralf

[radioz]

It seems to me that it is a dangerous design to have the landing zone on /. This virtually guarantees problems when there is not enough space to hold uploaded files.

I would think that it would be better if the default landing zone were located somewhere else (seperate partition maybe?).

[dinomic]

It seems to me that it is a dangerous design to have the landing zone on /. This virtually guarantees problems when there is not enough space to hold uploaded files.

I would think that it would be better if the default landing zone were located somewhere else (seperate partition maybe?).

I think radioz is right! I may have just run into that problem myself. Here's the scenario...
I currently have 3 drives:
1 x 60GB SSD (system drive, NOT in pool)
2 x 3TB (in pool)

I just tried to copy over about 150GB to the Storage Pool and was shocked to see that the System Drive was filling up. However, as I was seeing the files slowly appearing on the Pool Drives, I thought that the ones on the System Drive would be removed as soon as they were balanced on the Pool Drives. However, that doesn't seem to be the case. Whilst monitoring all locations with File Browser, I noticed that the files which had already been copied to the Pool Drives were showing up with an arrow on the icon on the System Drive, but they weren't being deleted? Do they only get deleted when the whole copy operation has been completed? In which case, with only 60GB on my System Drive, I'm limited to only copying 50Gb-odd at a time.

The machine is now in a "broken" state - web-based access to the HDA just gives Ruby On Rails errors, and any options I run on the "greyhole" command return the following error:
Can't describe tasks with query: DESCRIBE tasks - Error : Got error 28 from the Storage Engine

[gboudreau]

The files with an arrow are not actual files, they are symbolic links. Those will always be there, as they are what allows Samba to show files in your shares.
Those links are very small, and having them on your / partition costs almost nothing.
So files that are moved into the pool are not using space on your / partition, unless you added the / partition into the pool.