I have a 1TB Samsung 840 EVO as my Cache drive on one of my unRAID servers. It has became slow it seems... If I transfer to it, I get ~105MB/s (capped by network) but that quickly goes down to about 20-30MB/s. If I transfer to my other unRAID server that uses a WD Black drive, I get a solid 105MB/s the entire time. I'm showing no SMART errors on the SSD itself. I'm using XFS as the filesystem.

 

I run Sonarr/Sabnzbd on it 24/7, so perhaps it just needs the trim command sent to it. Is there a way to do this on unRAID?

I have a 1TB Samsung 840 EVO as my Cache drive on one of my unRAID servers. It has became slow it seems... If I transfer to it, I get ~105MB/s (capped by network) but that quickly goes down to about 20-30MB/s. If I transfer to my other unRAID server that uses a WD Black drive, I get a solid 105MB/s the entire time. I'm showing no SMART errors on the SSD itself. I'm using XFS as the filesystem.

 

I run Sonarr/Sabnzbd on it 24/7, so perhaps it just needs the trim command sent to it. Is there a way to do this on unRAID?

 

Does sabnzb do the download/repair/unpack on that drive? So many written and deleted files on that disk? Wonder how long the drive will last. Please keep me posted.

I have a 1TB Samsung 840 EVO as my Cache drive on one of my unRAID servers. It has became slow it seems... If I transfer to it, I get ~105MB/s (capped by network) but that quickly goes down to about 20-30MB/s. If I transfer to my other unRAID server that uses a WD Black drive, I get a solid 105MB/s the entire time. I'm showing no SMART errors on the SSD itself. I'm using XFS as the filesystem.

 

I run Sonarr/Sabnzbd on it 24/7, so perhaps it just needs the trim command sent to it. Is there a way to do this on unRAID?

 

Does sabnzb do the download/repair/unpack on that drive? So many written and deleted files on that disk? Wonder how long the drive will last. Please keep me posted.

 

Yes it does but due to being a 1TB drive, it maximum writes are insane. It'd take 1000TB of written data to reach the warrantied 1000 cycles, but many of these drives get closer to 2500 cycles based on tests people have done. I really don't see 2500TB of data being push through the drive no matter how long I use it. It's been in there for a year and it's only got 44 cycles, or ~44TB wrote.

 

I use two 1TB SSDs on my Windows PC for transferring and converting blu-rays to MKV, which is about ~80GB of writes per movie between them. Never had a problem, each of them sitting at about 100TB wrote (1/10th what they are warrantied for) after ~2 years of using them. 0 reallocated or pending sectors on either.

 

If you hooked up some smaller SSD, you might run into write problems.

I have a 1TB Samsung 840 EVO as my Cache drive on one of my unRAID servers. It has became slow it seems... If I transfer to it, I get ~105MB/s (capped by network) but that quickly goes down to about 20-30MB/s. If I transfer to my other unRAID server that uses a WD Black drive, I get a solid 105MB/s the entire time. I'm showing no SMART errors on the SSD itself. I'm using XFS as the filesystem.

 

I run Sonarr/Sabnzbd on it 24/7, so perhaps it just needs the trim command sent to it. Is there a way to do this on unRAID?

 

Does sabnzb do the download/repair/unpack on that drive? So many written and deleted files on that disk? Wonder how long the drive will last. Please keep me posted.

 

Yes it does but due to being a 1TB drive, it maximum writes are insane. It'd take 1000TB of written data to reach the warrantied 1000 cycles, but many of these drives get closer to 2500 cycles based on tests people have done. I really don't see 2500TB of data being push through the drive no matter how long I use it. It's been in there for a year and it's only got 44 cycles, or ~44TB wrote.

 

I use two 1TB SSDs on my Windows PC for transferring and converting blu-rays to MKV, which is about ~80GB of writes per movie between them. Never had a problem, they each are sitting at around 250TB wrote.

 

If you hooked up some smaller SSD, you might run into write problems.

 

Nice!

 

Would it be possible to configure sabnzb/unraid so that files get downloaded and repaired on one SSD and then unpacked on a 2nd SSD? The files will get moved to the array later I presume?

I have a 1TB Samsung 840 EVO as my Cache drive on one of my unRAID servers. It has became slow it seems... If I transfer to it, I get ~105MB/s (capped by network) but that quickly goes down to about 20-30MB/s. If I transfer to my other unRAID server that uses a WD Black drive, I get a solid 105MB/s the entire time. I'm showing no SMART errors on the SSD itself. I'm using XFS as the filesystem.

 

I run Sonarr/Sabnzbd on it 24/7, so perhaps it just needs the trim command sent to it. Is there a way to do this on unRAID?

 

Does sabnzb do the download/repair/unpack on that drive? So many written and deleted files on that disk? Wonder how long the drive will last. Please keep me posted.

 

Yes it does but due to being a 1TB drive, it maximum writes are insane. It'd take 1000TB of written data to reach the warrantied 1000 cycles, but many of these drives get closer to 2500 cycles based on tests people have done. I really don't see 2500TB of data being push through the drive no matter how long I use it. It's been in there for a year and it's only got 44 cycles, or ~44TB wrote.

 

I use two 1TB SSDs on my Windows PC for transferring and converting blu-rays to MKV, which is about ~80GB of writes per movie between them. Never had a problem, they each are sitting at around 250TB wrote.

 

If you hooked up some smaller SSD, you might run into write problems.

 

Nice!

 

Would it be possible to configure sabnzb/unraid so that files get downloaded and repaired on one SSD and then unpacked on a 2nd SSD? The files will get moved to the array later I presume?

 

Not sure, 1 SSD seems to do it fast enough even when dealing with huge 50GB NZBs.

Fixed it by installing the trim plugin here:

http://lime-technology.com/forum/index.php?topic=36543.0

 

Then I telnetted in and manually invoked the trim command.

I have a 1TB Samsung 840 EVO as my Cache drive on one of my unRAID servers. It has became slow it seems... If I transfer to it, I get ~105MB/s (capped by network) but that quickly goes down to about 20-30MB/s. If I transfer to my other unRAID server that uses a WD Black drive, I get a solid 105MB/s the entire time. I'm showing no SMART errors on the SSD itself. I'm using XFS as the filesystem.

 

I run Sonarr/Sabnzbd on it 24/7, so perhaps it just needs the trim command sent to it. Is there a way to do this on unRAID?

You might also see a problem I noticed with the 840 EVO. If I throw a lot of data on to it (20 – 30 GB) the temperature increases pretty quickly and after a while the drive is throttling and speed drops in order for the drive not to get damaged.

 

If you change Tunable (poll_attributes) to some low numbers you might be able to notice it.

I have a 1TB Samsung 840 EVO as my Cache drive on one of my unRAID servers. It has became slow it seems... If I transfer to it, I get ~105MB/s (capped by network) but that quickly goes down to about 20-30MB/s. If I transfer to my other unRAID server that uses a WD Black drive, I get a solid 105MB/s the entire time. I'm showing no SMART errors on the SSD itself. I'm using XFS as the filesystem.

 

I run Sonarr/Sabnzbd on it 24/7, so perhaps it just needs the trim command sent to it. Is there a way to do this on unRAID?

You might also see a problem I noticed with the 840 EVO. If I throw a lot of data on to it (20 – 30 GB) the temperature increases pretty quickly and after a while the drive is throttling and speed drops in order for the drive not to get damaged.

 

If you change Tunable (poll_attributes) to some low numbers you might be able to notice it.

 

Luckily, hasn't happened for me. I have a 120mm fan right in front of it. Though upon inspecting my unRAID setup, a 3 year old 3TB drive just threw out 3 pending sectors. Ugh. Really don't have the $ to replace it right now.