Excessive Writes on Cache Drive (Solved)

[CybranNakh]

Hello everyone! 

 

My server has been up for 11 days now... during that time there have been 55 million writes to my cache drive. Now as much as I would like to think my cache is doing its job... these seems rather excessive and the perfect way to kill the drive. 

 

How can I tell of this is part of the BTRFS format bug rather than something I have done incorrectly? Should I just try and switch my Cache format?

 

I am running a Plex server which I have read on some of the related forums can cause larger cache writing but not this much to my knowledge...Thanks for any help! 

 

Solution:

My array was encrypted XFS while the cache was btrfs which others have said makes the Loop2 bug worse. My solution was to convert the cache to encrypted xfs.... for me this has solved the excessive writes back down to normal levels! 

 

apollo-diagnostics-20200706-1608.zip

Edited July 14, 2020 by CybranNakh
solution

[JorgeB]

Try this, also look at the SMART to calculate the GBs written per day, number of writes in the GUI doesn't mean much.

[CybranNakh]

I have tried that command. Thank you for your reply! How would I look at the SMART? From googling it looks like I can just download the SMART Report? I have attached it here (removed serial number). 

SMART Report.txt

[JorgeB]

Check the thread linked above, there's even a scrip to calculate that for you.

[CybranNakh]

On 7/6/2020 at 11:55 PM, johnnie.black said:

Try this, also look at the SMART to calculate the GBs written per day, number of writes in the GUI doesn't mean much.

This helped alot! While the commands did not work for me, I found another comment in the thread discussing the Loop2 error and encrypted xfs for the array and btrfs on the cache. The solution for me was to convert the cache drive to encrypted xfs. So far, the writes and GBs written have returned to normal levels! (This triggered my memory of this cache issue coinciding with converting my array to encrypted xfs.)