swi1

updated

12 years ago

S

swi1

1 Rookie

•

33 Posts

0

7201

December 3rd, 2014 14:00

Is there a way to see file size and file location in specific node pool

Hi, We have two node pools and one of the pool reached 100% and other one has space. Now the smart-pool job is running from some time. Mean while - is there any command or option where we can see the file location (In which pool it resides) and the exact file size on specific node pool which it is being written.

  • AndrewChung

    132 Posts

    5253

    1

    Posted December 4th, 2014 01:00

    With a single -D you will only get the primary LIN (logical inode) information.  If you want to see where the blocks physically live you need to use -DD.

    What you highlighted shows you where the actual inodes reside.  In your particular case:

    1,0,67930112:512 = Node ID 1, Drive Lnum 0, LBA 67930112, 512 bytes

    1,6,237288448:512 = Node ID 1, Drive Lnum 6, LBA 237288448, 512 bytes

    2,7,134513152:512 = Node ID 2, Drive Lnum 7, LBA 134513152, 512 bytes

    2,21,42245120:512 = Node ID 2, Drive Lnum 2, LBA 42245120, 512 bytes


    The NodeID and the logical node number are not the same.  The NodeID is a monotonically increasing number that you cannot change.  It is set in the order that you add nodes to a cluster.

    The Drive Bay number is NOT the same as what you would expect.  This is the physical bay number.  You can obtain the actual bay number by doing a isi devices.  The Lnum value is the one you want to reference.

    LBA and bytes are pretty straight forward.


    When you do a -DD you will get a large block list of where the data sits.  It is in the same format as the inodes except you will see an additional field that looks like: #16 or #1.  That is the count of how many contiguous blocks starting at the given LBA are in use.  Our stripe is normally 16 blocks so the largest value you will see in this field is #16

  • 5253

    1

    Posted December 3rd, 2014 16:00

    Hello Swi,

    you can use the isi get -D command to discover details about a file like where they live and on what pool. it may be helpful to grep what you are looking for as the output is rather large.

    Example:

    Mr-Cluster-1# isi get -D mc
    POLICY  W  LEVEL PERFORMANCE COAL  ENCODING      FILE              IADDRS
    default   4x/2 concurrency on    N/A           ./                <1,0,67930112:512>, <1,6,237288448:512>, <2,7,134513152:512>, <2,21,42245120:512> ct: 1417651615 rt: 1417651757
    *************************************************
    * IFS inode: [ 1,0,67930112:512, 1,6,237288448:512, 2,7,134513152:512, 2,21,42245120:512 ]
    *************************************************
    *
    *  Inode Version:      6
    *  Dir Version:        2
    *  Inode Revision:     2
    *  Inode Mirror Count: 4
    *  Recovered Flag:     0
    *  Recovered Groups:   0
    *  Link Count:         2
    *  Size:               0
    *  Mode:               040755
    *  Flags:              0xe0
    *  Physical Blocks:    0
    *  LIN:                1:0006:43e0
    *  Logical Size:       None
    *  Shadow refs:        0
    *  Do not dedupe:      0
    *  Restripe Cursor:    None
    *  Abort Cursor:       None
    *  Layout cursor:      None
    *  Last Modified:      1417651615.545135254
    *  Last Inode Change:  1417651757.233943311
    *  Create Time:        1417651615.545135254
    *  Rename Time:        1417651757
    *  Write Caching:      Enabled
    *  Parent Lin          1:0001:0085
    *  Parent Hash:        658198
    *  Snapshot IDs:       None
    *  Last Paint ID:          1
    *  Domain IDs:         None
    *  LIN needs repair:   False
    *  Manually Manage:
    *       Access         False
    *       Protection     False
    *  Protection Policy:  Diskpool default
    *  Current Protection: 4x
    *  Future Protection:  0x
    *  Disk pools:         policy any pool group ID -> data target bootstrap pool ID, metadata target bootstrap pool ID
    *  SSD Strategy:       metadata
    *  SSD Status:         complete
    *  Layout drive count: 0
    *  Access pattern: 0
    *  Data Width Device List:
    *  Meta Width Device List:
    *
    *  File Data (0 bytes):
    *    Metatree Depth: 0
    *
    *  Dynamic Attributes (42 bytes):
    *
            ATTRIBUTE                OFFSET SIZE
            Rename time              0      5
            New file attribute       5      23
            Disk pool policy ID      28     5
            Last snapshot paint time 33     9
    *
    *************************************************

    *  NEW FILE ATTRIBUTES
    *  Access attributes:  active
    *  Write Cache:  on
    *  Access Pattern:  concurrency
    *  At_r: 0
    *  Protection attributes:  active
    *  Protection Policy:  Diskpool default
    *  Disk pools:         policy any pool group ID
    *  SSD Strategy:       metadata
    *
    *************************************************

    *
    *************************************************

    default   4x/2 concurrency on    N/A           ../               <1,1,168595968:512>, <1,9,236748800:512>, <2,8,181184512:512>, <2,17,67410432:512> ct: 1417651682 rt: 0
    *************************************************
    * IFS inode: [ 1,1,168595968:512, 1,9,236748800:512, 2,8,181184512:512, 2,17,67410432:512 ]
    *************************************************
    *
    *  Inode Version:      6
    *  Dir Version:        2
    *  Inode Revision:     2
    *  Inode Mirror Count: 4
    *  Recovered Flag:     0
    *  Recovered Groups:   0
    *  Link Count:         3
    *  Size:               20
    *  Mode:               040755
    *  Flags:              0xe0
    *  Physical Blocks:    0
    *  LIN:                1:0001:0085
    *  Logical Size:       None
    *  Shadow refs:        0
    *  Do not dedupe:      0
    *  Restripe Cursor:    None
    *  Abort Cursor:       None
    *  Layout cursor:      None
    *  Last Modified:      1417651757.233943311
    *  Last Inode Change:  1417651757.233943311
    *  Create Time:        1417651682.024569503
    *  Rename Time:        0
    *  Write Caching:      Enabled
    *  Parent Lin          2
    *  Parent Hash:        848367
    *  Snapshot IDs:       None
    *  Last Paint ID:          1
    *  Domain IDs:         None
    *  LIN needs repair:   False
    *  Manually Manage:
    *       Access         False
    *       Protection     False
    *  Protection Policy:  Diskpool default
    *  Current Protection: 4x
    *  Future Protection:  0x
    *  Disk pools:         policy any pool group ID -> data target bootstrap pool ID, metadata target bootstrap pool ID
    *  SSD Strategy:       metadata
    *  SSD Status:         complete
    *  Layout drive count: 0
    *  Access pattern: 0
    *  Data Width Device List:
    *  Meta Width Device List:
    *
    *  File Data (34 bytes):
    *    Metatree Depth: 1
    *
    *  Dynamic Attributes (37 bytes):
    *
            ATTRIBUTE                OFFSET SIZE
            New file attribute       0      23
            Disk pool policy ID      23     5
            Last snapshot paint time 28     9
    *
    *************************************************

    *  NEW FILE ATTRIBUTES
    *  Access attributes:  active
    *  Write Cache:  on
    *  Access Pattern:  concurrency
    *  At_r: 0
    *  Protection attributes:  active
    *  Protection Policy:  Diskpool default
    *  Disk pools:         policy any pool group ID
    *  SSD Strategy:       metadata
    *
    *************************************************

    *
    *************************************************

  • swi1

    1 Rookie

    •

    33 Posts

    5253

    1

    Posted December 4th, 2014 00:00

    Hi tracy,

    I think this  command is not showing the size on specific pool, correct me I am wrong.

    Could you brief me what its saying about  the result below----------------------

    * IFS inode: [ 1,0,67930112:512, 1,6,237288448:512, 2,7,134513152:512, 2,21,42245120:512 ]

    *************************************************

  • Peter_Sero

    6 Operator

    •

    1169 Posts

    5253

    1

    Posted December 4th, 2014 06:00

    Quickest way is to look at the lines of this kind within the isi get -D output:

    *  Disk pools:         policy any pool group ID -> data target bootstrap pool ID, metadata target bootstrap pool ID


    (For files just half way being migrated this is either the old or the new pool,

    whereas counting the blocks from the  -DD output will give precise results.)

  • swi1

    1 Rookie

    •

    33 Posts

    5253

    0

    Posted December 4th, 2014 10:00

    Ok, related to this topic another question...Suppose we have two node pools(NL and X) say Node 1-5 are x Node pool and 6-10 are NL node pool. If a client is connected to  node 6 and starts writing a new file, does the data and parity is striped over nodes 1-10 or it uses only x node pool? - The one FS version is 7.0.XX

  • Peter_Sero

    6 Operator

    •

    1169 Posts

    5253

    1

    Posted December 4th, 2014 20:00

    Files go either to the x OR to the nl pool, as determined in your filepool rules (aka Smartpools policies),

    but never will be striped as "9+1" (actually 18+2/2) in your example.

    With spill-over enabled and one pool running full, some stripe units of the same file

    can be on x (8+2/2) and others on nl (8+2/2), but such a situation should not be

    considered as permanent. Same for a file while it is being migrated from one pool to another.

    The filepool rules can include any combination of directory paths, file name patterns,

    file size and age etc. but NOT the identity of the node which creates a file.

    The catch is that nodes and paths can be associated at two different stages:

    1. filepool rules:  path /ifs/whatever -> node pool or tier x (or nl or...)

    2. access zones: subnet:adressspool <-> nodes:interfaces <-> root path of access zone

    Just make sure that the filepool rules and access zone definitions

    are consistent with respect to nodes and paths, if you want data

    arriving at one node pool x also getting written on this node pool.

    hth

    -- Peter

  • swi1

    1 Rookie

    •

    33 Posts

    5253

    0

    Posted December 4th, 2014 20:00

    I gotcha Peter, because of spillover the data may reside anywhere. If the node pools or tiers are free,It should be on single node or tier - according to file pool policy.Am I right?

  • Peter_Sero

    6 Operator

    •

    1169 Posts

    5253

    0

    Posted December 4th, 2014 21:00

    Yes ("single node /pool/ or tier", that is. "single node" is impossible).