slab/kernel memory problem

From: Alexander Lazarevich (alazarev_at_itg.uiuc.edu)
Date: 09/29/05

  • Next message: Bharadwaj: "questions related with partion ?"
    Date: Thu, 29 Sep 2005 11:23:26 -0500 (CDT)
    To: General Red Hat Linux discussion list <redhat-list@redhat.com>
    
    

    We have a RHEL4-AS 64bit system, and a FC364bit system. Both running most
    recent kernels.

    We've got bad memory problems. The kernel is sucking up quite a bit of
    memory, check out Slab in meminfo:

    [root@hostname ~]# cat /proc/meminfo
    MemTotal: 8078596 kB
    MemFree: 486480 kB
    Buffers: 1075532 kB
    Cached: 527284 kB
    SwapCached: 1384 kB
    Active: 2120352 kB
    Inactive: 1019628 kB
    HighTotal: 0 kB
    HighFree: 0 kB
    LowTotal: 8078596 kB
    LowFree: 486480 kB
    SwapTotal: 2096472 kB
    SwapFree: 2053372 kB
    Dirty: 28 kB
    Writeback: 0 kB
    Mapped: 1564776 kB
    Slab: 4424180 kB
    CommitLimit: 6135768 kB
    Committed_AS: 1836712 kB
    PageTables: 9952 kB
    VmallocTotal: 34359738367 kB
    VmallocUsed: 1356 kB
    VmallocChunk: 34359736951 kB
    HugePages_Total: 0
    HugePages_Free: 0
    Hugepagesize: 2048 kB

    Then, slabtop is showing what is using all that slab mem:

      Active / Total Objects (% used) : 9028210 / 9033666 (99.9%)
      Active / Total Slabs (% used) : 1105829 / 1105834 (100.0%)
      Active / Total Caches (% used) : 94 / 135 (69.6%)
      Active / Total Size (% used) : 108142.02K / 108819.82K (99.4%)
      Minimum / Average / Maximum Object : 0.02K / 0.01K / 128.00K

       OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME
    4181455 4181452 99% 0.78K 836291 5 3345164K ext3_inode_cache
    4129011 4129007 99% 0.23K 242883 17 971532K dentry_cache

    Why in the heck is the system trying to use all this memory to cache
    ext3 inodes? This happens on both FC364bit and RHEL4-AS64bit.

    I've search google and bugzilla, couldn't find anything.

    Anyone see this as well?

    Thanks!

    Alex

    -- 
    redhat-list mailing list
    unsubscribe mailto:redhat-list-request@redhat.com?subject=unsubscribe
    https://www.redhat.com/mailman/listinfo/redhat-list
    

  • Next message: Bharadwaj: "questions related with partion ?"

    Relevant Pages

    • Re: [BUG] New Kernel Bugs
      ... The big difference with large mmap'ed files is that later kernels will ... actually track dirty ratios for dirty mmap'ed pages. ... So in older kernels, you can dirty as much memory as you want, and the ... First dbclean creates a new dcc_db file by copying from the old file. ...
      (Linux-Kernel)
    • Re: [PATCH 0/8][for -mm] mem_notify v6
      ... > Here's a test program that allocates memory and frees on notification. ... CommitLimit: 63804 kB ... VmallocChunk: 908976 kB ...
      (Linux-Kernel)
    • RE: la la la la ... swappiness
      ... memory out the way we want it, the only rogue player is the tar update ... desperate and not only set swappiness to zero but I've also tried setting ... VmallocChunk: 113584 kB ...
      (Linux-Kernel)
    • Re: Possible Bug in VM accounting (Committed_AS) on x86_64 architecture ?
      ... malloc() returns -ENOMEM. ... VmallocChunk: 536601679 kB ... It allocates memory of different sizes (not that it matters much due ... for 32 bit architecture? ...
      (Linux-Kernel)
    • Re: Still possible to directly boot without loader?
      ... Since BTX no longer has this limitation (and hasn't for a long ... remove the masking from boot2. ... This allows boot2 to load kernels ... The kernel is linked at 0xc0000000 but loade din low memory, ...
      (freebsd-stable)