Odd CPU meter behaviour EDIT - IOWAIT HDD choking problem


I am using Atom 2.0 GHz dual core dedicated server as my seedbox. No need to install xeon because only two users will be using it. All the meters on the main panel works as they should but CPU meter. If I log to the SSH and check cpiu usage it shows 4.5% when cpu meter shows about 44%. While downloading 100 MB/s rtorrent cpu usage in terminal shows 49%, cpu meter on the dashboard says 450%(?)

It’s weird, is this the way it should work?


I think I’ve found the issue. 90% of cpu is IOWAIT. Can quotas slow down the disk? It’s choking with only 20 MB/s of write.

If you’re speaking of the dedibox sc/xc, then disk io is definitely in short supply. It’s a single 5200rpm 2.5 inch drive. Torrents utilize random io operations, which tend to be far more straining on the drive than sequential operations. The quotas themselves shouldn’t cause it, but if you have multiple torrents or multiple users accessing the drive then slowdowns may be expected.

1 Like

So I was cheated because I can’t even use 1 Gbps network supplied because disk won’t be fast enough. Good to know.

Is there any way I can improve the situation without buying new dedicated server?

Unfortunately, not a whole lot can be done. You may see an improvement if you alter your usage patterns though. The RAM is the fastest storage medium on the server. So, if your intent is solely to build buffer, avoid downloading items that are larger than your total ram (ie. stick to tv shows and small movies). It’s going to be the flushing of data to disk that absolutely kills the upload speed and further the responsiveness of the server. Also avoid downloading too many items at once - try to keep it to one, maybe two. Bear in mind that background processes such as Plex transcoding or any disk intensive apps are going to mess with your IO even further.

What about cache settings, can it improve big files write speed? Im tied to this server for months so I need to use it anyway, even if it lags.

I checked smart data to be sure and disk is brand new but with HDPARM for rotation rate I got 7200 RPM not 5400 so it should be much faster than 20 MB/s. I’ll try reinstalling the system and trying deluge instead of rtorrent.

This is the model: HGST HTE721010A9E630

Test - http://www.tomshardware.com/reviews/travelstar-7k1000-hdd-review,3479-3.html

I can say more. I have it’s direct competitor currently installed in laptop - WD7500BPKT and it’s a way faster. There’s clearly something wrong with the configuration or system (Debian 8).

When I installed the quickbox with quotas I had to set first quota manually by using setdisk for master user, otherwise it showed -1 as space. Don’t know if it was intentional to leave it unset after the install.


Got reponse from hosting. It’s hardware problem and it’s currently being fixed.

What is the server you are using, and at which provider? Curious as I’m fairly positive Online.net (the larger reseller of Atom servers) doesn’t use 7200rpm drives in their configurations.


Im using Dedibox-SC at Online.net.

Here’s full data of the hdparm:


ATA device, with non-removable media
        Model Number:       HGST HTE721010A9E630
        Serial Number:     XXXXXXXXX
        Firmware Revision:  JB0OA3M0
        Transport:          Serial, ATA8-AST, SATA 1.0a, SATA II Extensions, SATA Rev 2.5, SATA Rev 2.6, SATA Rev 3.0; Revision: ATA8-AST T13 Project D1697 Revision 0b
        Used: unknown (minor revision code 0x0028)
        Supported: 8 7 6 5
        Likely used: 8
        Logical         max     current
        cylinders       16383   16383
        heads           16      16
        sectors/track   63      63
        CHS current addressable sectors:   16514064
        LBA    user addressable sectors:  268435455
        LBA48  user addressable sectors: 1953525168
        Logical  Sector size:                   512 bytes
        Physical Sector size:                  4096 bytes
        Logical Sector-0 offset:                  0 bytes
        device size with M = 1024*1024:      953869 MBytes
        device size with M = 1000*1000:     1000204 MBytes (1000 GB)
        cache/buffer size  = unknown
        Form Factor: 2.5 inch
        Nominal Media Rotation Rate: 7200
        LBA, IORDY(can be disabled)
        Queue depth: 32
        Standby timer values: spec'd by Standard, no device specific minimum
        R/W multiple sector transfer: Max = 16  Current = 16
        Advanced power management level: 254
        DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 udma5 *udma6
             Cycle time: min=120ns recommended=120ns
        PIO: pio0 pio1 pio2 pio3 pio4
             Cycle time: no flow control=120ns  IORDY flow control=120ns
        Enabled Supported:
           *    SMART feature set
                Security Mode feature set
           *    Power Management feature set
           *    Write cache
           *    Look-ahead
           *    Host Protected Area feature set
           *    WRITE_BUFFER command
           *    READ_BUFFER command
           *    NOP cmd
           *    DOWNLOAD_MICROCODE
           *    Advanced Power Management feature set
                Power-Up In Standby feature set
           *    SET_FEATURES required to spinup after power up
                SET_MAX security extension
           *    48-bit Address feature set
           *    Device Configuration Overlay feature set
           *    Mandatory FLUSH_CACHE
           *    FLUSH_CACHE_EXT
           *    SMART error logging
           *    SMART self-test
           *    General Purpose Logging feature set
           *    WRITE_{DMA|MULTIPLE}_FUA_EXT
           *    64-bit World wide name
           *    IDLE_IMMEDIATE with UNLOAD
           *    WRITE_UNCORRECTABLE_EXT command
           *    {READ,WRITE}_DMA_EXT_GPL commands
           *    Segmented DOWNLOAD_MICROCODE
           *    Gen1 signaling speed (1.5Gb/s)
           *    Gen2 signaling speed (3.0Gb/s)
           *    Gen3 signaling speed (6.0Gb/s)
           *    Native Command Queueing (NCQ)
           *    Host-initiated interface power management
           *    Phy event counters
           *    NCQ priority information
           *    READ_LOG_DMA_EXT equivalent to READ_LOG_EXT
                Non-Zero buffer offsets in DMA Setup FIS
           *    DMA Setup Auto-Activate optimization
                Device-initiated interface power management
                In-order data delivery
           *    Software settings preservation
           *    SMART Command Transport (SCT) feature set
           *    SCT Write Same (AC2)
           *    SCT Error Recovery Control (AC3)
           *    SCT Features Control (AC4)
           *    SCT Data Tables (AC5)
        Master password revision code = 65534
        not     enabled
        not     locked
        not     expired: security count
                supported: enhanced erase
Checksum: correct

Interesting. Is this the 2016 or 2015 variant? I don’t have anything to compare it to - the only XC machine I have on hand is the SSD variant.

The 2016 variants are notoriously bad due to “BIOS problems” as their motherboard is a custom in-house build.

Edit: Though I will just mention the the processor on the SC variant is a slouch - last time I did an installation on it, it clocked in around 2 hours. Most quickbox installs are done in 20-30 minutes.

This is 2016 variant. Write speed may be somehow related to these “BIOS problems” you mentioned.

It’s being fixed now, they are starting from rack 20 and im on rack 33. I need to say I also got good technical assistance and even some recompense for this problem so I definitely recommend.

Maybe they only put 5400 rpm in 2015 variants and after a lot of support tickets they decided to put faster drives in next model.

I hope things work out for you :smiley:

1 Like

Thanks! And thanks for the fast response too! I hope quickbox script won’t choke after the fix.

EDIT:: ah didn’t saw your edit. Installation got about 40 minutes on SC.


Hey Excelsior,

I too am having exactly the same issue…

100% cpu and above all the time in quickbox. Its running like a slug… and its only been the past couple of weeks this bad.

I’m online.net as well with a dedibox SC. I would love some help on this or how you approached support to find/get a resolution.

All the best and thanks.

Hey, @Hornet, Excelsior and Liara helped me with that too,
Run this command it will help you to know how far your HDD is in the grave.

dd if=/dev/zero of=$HOME/test.img bs=1G count=1 oflag=direct

It should return you something like that

1+0 records in
1+0 records out
1073741824 bytes (1.1 GB, 1.0 GiB) copied, 137.002 s, 7.8 MB/s

It’s your HDD transfert speed, a normal HDD goes around 100 MB/s (ie 119 on mine)
To approch Assitance, call them or create a ticket, telling that your server is anormally slow and include your test.

They probably purpose you an XC2015 at a reduced price as they did for both of us.

Hope it’ll help you

Hey @samnawak ,

Thanks heaps for your help. I’m currently running this test now to see. Will post results when done.

I started a ticket also and they have confirmed it is related to the current service problem #720. I asked how long it would take or if they had a projected timeline for resolution.

They had no idea when it will be resolved and i should check back regularly for updates on the problem. :frowning: boo!

I’m going to post them my results and ask them about a faster solution/new server option.

May i ask what ball park figure i should be expecting to pay for a xc2015?

Thanks again for the help. :smiley:

No problems, we are all here to help each other,

They took about 24h to diagnose and deliver me the XC2015,
Price is 11€HT/month for me,

Hope it will solve your problems :wink:

1 Like

Thanks for the info @samnawak .

I’m currently talking to support to see what they can offer.

Looks like they could be pointing me towards a XC 2015. Although it does look like it is actually a 2014 with better specs then the 2016 but a slower connection speed / slower guaranteed speed.

Bit of a coin toss on whether i wait it out and see if they can fix the problem to keep the faster connection or just get a working server again.


I also got XC 2015 as replacement. Server is much, much faster than SC2016 with only a little big increased price. ID is LIMITED (so it’s a limited offer). Network is still 1Gbps. I recommend accepting that. HDD is 7200 RPM - HGST HTS721010A9E630

8 core 2.6 GHz Atom is in multi-core operations near Xeon E3! It’s a excellent offer mate.

Thanks for the hot tip @Excelsior

i think i just needed to bounce the idea around and see what any one else thought.

Appreciate the info an I’ll let ya know how it goes.

And all is complete… new server in and all setup again.

You were on the money @Excelsior its much faster.

Thanks to everyone who helped out.

All the best.

Glad I’ve seen this thread. My 2016 SC 1TB 2core/4gb 8.99€ was slow as hell.

I was offered replacement 2015 model C2750 8core /8gb ram/1tb for 10.99€ .

It’s 2€ more, but performs way better.

I tried running this command on old server
dd bs=1M count=64 if=/dev/zero of=test conv=fdatasync
and results I had were in range of 9-13 MB/s . Very bad!