Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Economical storage vps immune from days long fsck?
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Economical storage vps immune from days long fsck?

For the second time this year, I am waiting days for my ServerHub hosted vps to come back because something happened and they had to run fsck on their enormous array.

This is intolerable. Can you recommend someone who does this right and doesn't charge an arm and a leg? I need a fair bit of storage, half-terabyte at least. I one 1TB at Serverhub. I don't need vast compute or data transfer.

«1

Comments

  • mikhomikho Member, Host Rep

    If OpenVZ NAT is ok, I can offer pretty cheap storage in 500gb chunks.

  • mikho said: If OpenVZ NAT is ok, I can offer pretty cheap storage in 500gb chunks

    I clicked your link for Lenoir storage VPS but it looked like 500gb was the maximum size rather than the minimum.

    @Francisco what happens if your slice cluster needs an fsck or rebuild?

    @subhacker you should really have a 2nd copy of the data at another location, of course. But maybe you want a dedi like a KS1 if you can find one. That's 500GB at $5/month with no RAID if you're ok with that.

  • Get a SyS Storage server while it lasts - good for cold backups and the such. 2TB for $6/month, this ARM won't cost you a leg

    Thanked by 2eol dahartigan
  • mikhomikho Member, Host Rep

    @willie said:

    mikho said: If OpenVZ NAT is ok, I can offer pretty cheap storage in 500gb chunks

    I clicked your link for Lenoir storage VPS but it looked like 500gb was the maximum size rather than the minimum.

    That is correct. Using multiple volumes to avoid having the complete array to fsck.
    The decision was made to offer rather large storage options without making them to large and hard to handle.

    Thanked by 1vimalware
  • Shot2 said: Get a SyS Storage server

    I would recommend against this idea. This "server" comes with a single 2Tb drive, better get storage VPS from @UltraVPS, @SpeedyKVM or @Abdullah (aka HostHatch). This will provide you with at least minimal redundancy on the hardware level.

    Above-mentioned providers are welcome to join in and rap about high level of reliability of their setups.

    Thanked by 1UltraVPS
  • Presumably it depends on what filesystem is used.

    Whilst BTRFS/ZFS have scrubs I believe they are done online.

  • subhacker said: For the second time this year, I am waiting days for my ServerHub hosted vps to come back because something happened and they had to run fsck on their enormous array.

    I've seen 2-3 weeks for a fsck to later find out its a RAID card issue.

  • subhacker said: For the second time this year, I am waiting days for my ServerHub hosted vps to come back because something happened and they had to run fsck on their enormous array.

    I also have a VPS with them in Phoenix in the same situation. Ridiculous. What are they doing losing power in the first place.

  • @subhacker said:
    For the second time this year, I am waiting days for my ServerHub hosted vps to come back because something happened and they had to run fsck on their enormous array.

    This is intolerable. Can you recommend someone who does this right and doesn't charge an arm and a leg? I need a fair bit of storage, half-terabyte at least. I one 1TB at Serverhub. I don't need vast compute or data transfer.

    i had the same situation with them, more than one week with an fsck, the only reason i'm still with them is because they look like a reliable company(zxhost panic).

    regards!

  • This nVme+ 500GB/1000GB bundle for $3.5/$7 is the better deal. (need to wait for re stock right now)

    https://www.lowendtalk.com/discussion/152637/new-nvme-block-storage

    Thanked by 1letbox
  • ehabehab Member
    edited November 2018

    @key900 you know what to do.

    Thanked by 1letbox
  • @mikho said:
    If OpenVZ NAT is ok, I can offer pretty cheap storage in 500gb chunks.

    I think the 500GB chunks idea is a good one but I need a real IP.

  • @willie said:
    @subhacker you should really have a 2nd copy of the data at another location, of course. But maybe you want a dedi like a KS1 if you can find one. That's 500GB at $5/month with no RAID if you're ok with that.

    It is a storage VPS in that it needs a good chunk of storage at low cost but it is not an archive or a backup. The data is highly dynamic. The bulk of the storage needs comes from the news server, which maybe I should finally dump. I'd rather not, though.

  • letboxletbox Member, Patron Provider
    edited November 2018

    Thanks you for mentioned us

    @vimalware said:

    This nVme+ 500GB/1000GB bundle for $3.5/$7 is the better deal. (need to wait for re stock right now)

    https://www.lowendtalk.com/discussion/152637/new-nvme-block-storage

    Those bundle will not come up anytime soon sadly.

    @ehab said:
    @key900 you know what to do.

    @subhacker said:
    For the second time this year, I am waiting days for my ServerHub hosted vps to come back because something happened and they had to run fsck on their enormous array.

    This is intolerable. Can you recommend someone who does this right and doesn't charge an arm and a leg? I need a fair bit of storage, half-terabyte at least. I one 1TB at Serverhub. I don't need vast compute or data transfer.

    Our Block Storage are come with Self healing and Triple ( 3 copy ) Data is replicated, making it fault tolerant.

  • mikho said: Using multiple volumes to avoid having the complete array to fsck.

    I don't see a way to order multiple volumes on a single VPS. Is there a way to do that? Thanks.

  • This sounds like a solution for a slice & slab from @Francisco... Had mine since public launch and can't reccomend it enough.

  • FranciscoFrancisco Top Host, Host Rep, Veteran

    @eastonch said:
    This sounds like a solution for a slice & slab from @Francisco... Had mine since public launch and can't reccomend it enough.

    Thanks :)

    I'd appreciate any/all feedback I can get on the product and the platform. Good and bad.

    Francisco

  • Francisco said: Get a slab!

    How fast is disk access? It it suitable for running a database?

  • @tetech said:

    Francisco said: Get a slab!

    How fast is disk access? It it suitable for running a database?

    It depends on the dataset and iops.

  • FranciscoFrancisco Top Host, Host Rep, Veteran

    @tetech said:

    Francisco said: Get a slab!

    How fast is disk access? It it suitable for running a database?

    It's awesome.

    If you need 10k iops that's a problem but if you're doing that much you're likely missing some key indexes anyway.

    Francisco

    Thanked by 1sin
  • sinsin Member
    edited November 2018

    @tetech said:

    Francisco said: Get a slab!

    How fast is disk access? It it suitable for running a database?

    It's very fast. I plan on storing all my Wordpress sites and databases on it (created a zfs mirror from 2x 256GB slabs..only $2.50 total which is crazy cheap).

    root@buyvmbsdsrv01:~ # diskinfo -cti /dev/da1
    /dev/da1
        512             # sectorsize
        274877906944    # mediasize in bytes (256G)
        536870912       # mediasize in sectors
        4096            # stripesize
        0               # stripeoffset
        33418           # Cylinders according to firmware.
        255             # Heads according to firmware.
        63              # Sectors according to firmware.
        BUYVM SLAB      # Disk descr.
        VOLUME-xxx      # Disk ident.
        Yes             # TRIM/UNMAP support
        Unknown         # Rotation rate in RPM
        Not_Zoned       # Zone Mode
    
    I/O command overhead:
        time to read 10MB block      0.074586 sec   =    0.004 msec/sector
        time to read 20480 sectors   2.574480 sec   =    0.126 msec/sector
        calculated command overhead         =    0.122 msec/sector
    
    Seek times:
        Full stroke:      250 iter in   0.013853 sec =    0.055 msec
        Half stroke:      250 iter in   0.014821 sec =    0.059 msec
        Quarter stroke:   500 iter in   0.030980 sec =    0.062 msec
        Short forward:    400 iter in   0.023949 sec =    0.060 msec
        Short backward:   400 iter in   0.026506 sec =    0.066 msec
        Seq outer:   2048 iter in   0.186547 sec =    0.091 msec
        Seq inner:   2048 iter in   0.177489 sec =    0.087 msec
    
    Transfer rates:
        outside:       102400 kbytes in   0.046892 sec =  2183741 kbytes/sec
        middle:        102400 kbytes in   0.047161 sec =  2171286 kbytes/sec
        inside:        102400 kbytes in   0.041751 sec =  2452636 kbytes/sec
    
    Asynchronous random reads:
        sectorsize:     54185 ops in    3.000584 sec =    18058 IOPS
        4 kbytes:       53456 ops in    3.000852 sec =    17814 IOPS
        32 kbytes:      51028 ops in    3.000726 sec =    17005 IOPS
        128 kbytes:     42272 ops in    3.000697 sec =    14087 IOPS
    
    Thanked by 3Francisco eol tetech
  • zkyez said: It depends on the dataset and iops.

    Sure, I was really after a qualitative comparing to a spinning HDD on the node, like failed for ServerHub in this case. Not hard numbers, but "about the same", "50% faster", etc.

  • NeoonNeoon Community Contributor, Veteran

    Well, get a 64GB SSD, the fsck will be fast.

    As long nothing goes unexpected, a wild fsck will not appear.

    Thanked by 1eol
  • sin said: It's very fast. I plan on storing all my Wordpress sites and databases on it (created a zfs mirror from 2x 256GB slabs..only $2.50 total which is crazy cheap).

    That's helpful, thanks.

  • @Neoon said:
    Well, get a 64GB SSD, the fsck will be fast.

    As long nothing goes unexpected, a wild fsck will not appear.

    Or a distributed filesystem and have the redundancy set as such that if a node does have to FSCK it doesn't matter.

  • @dragon2611 said:
    Or a distributed filesystem and have the redundancy set as such that if a node does have to FSCK it doesn't matter.

    That worked great for a certain provider..

  • Nah, use ZFS which does not need a file checker at all. A monthly scrub is good for everyone.

    Thanked by 1vimalware
Sign In or Register to comment.