VMware notes

From Noah.org
Revision as of 12:34, 23 July 2008 by Root (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search


VMware Linux kernel any-to-any patches

I always end up using these any-any patches for the Linux kernel. They are easy to apply.

   http://knihovny.cvut.cz/ftp/pub/vmware/readme.txt

ext3 tuning

VMware performance is helped a lot with some tweaks to ext3. See Disk Performance Tuning for more info.

VMware machine giving EXT3-fs error: Detected aborted journal

If you see the following in your /var/log/messages on a machine running VMware then this could be a VMware configuration issue:

Feb  7 09:42:14 roothack kernel: kswapd0: page allocation failure. order:0, mode:0x850
Feb  7 09:42:14 roothack kernel:
Feb  7 09:42:14 roothack kernel: Call Trace:<ffffffff801ea51a>{vgacon_cursor+0} <ffffffff80157b6e>{__alloc_pages+768}
Feb  7 09:42:14 roothack kernel:        <ffffffff8016c8fd>{alloc_page_interleave+61} <ffffffff80157c02>{__get_free_pages+11}
Feb  7 09:42:14 roothack kernel:        <ffffffff8015aa8c>{kmem_getpages+36} <ffffffff8015b227>{cache_alloc_refill+615}
Feb  7 09:42:14 roothack kernel:        <ffffffff8015aeef>{__kmalloc+123} <ffffffffa006ea38>{:jbd:__jbd_kmalloc+21}
Feb  7 09:42:14 roothack kernel:        <ffffffffa006a7b0>{:jbd:journal_get_undo_access+96}
Feb  7 09:42:14 roothack kernel:        <ffffffffa007baa9>{:ext3:ext3_try_to_allocate_with_rsv+84}
Feb  7 09:42:14 roothack kernel:        <ffffffffa007c1df>{:ext3:ext3_new_block+680} <ffffffffa007e396>{:ext3:ext3_alloc_block+7}
Feb  7 09:42:14 roothack kernel:        <ffffffffa007ff6a>{:ext3:ext3_get_block_handle+863}
Feb  7 09:42:14 roothack kernel:        <ffffffff801752f3>{__block_write_full_page+198} <ffffffffa00803bf>{:ext3:ext3_get_block+0}
Feb  7 09:42:14 roothack kernel:        <ffffffffa007eb26>{:ext3:ext3_ordered_writepage+245}
Feb  7 09:42:14 roothack kernel:        <ffffffff8015eada>{shrink_zone+3102} <ffffffff80157233>{__free_pages_ok+240}
Feb  7 09:42:14 roothack kernel:        <ffffffff8019c58e>{mb_cache_shrink_fn+94} <ffffffff801dcdf4>{__up_read+15}
Feb  7 09:42:14 roothack kernel:        <ffffffff8015f3b2>{balance_pgdat+506} <ffffffff8015f5fc>{kswapd+252}
Feb  7 09:42:14 roothack kernel:        <ffffffff80132ff0>{autoremove_wake_function+0} <ffffffff80130541>{finish_task_switch+55}
Feb  7 09:42:14 roothack kernel:        <ffffffff80132ff0>{autoremove_wake_function+0} <ffffffff80130590>{schedule_tail+11}
Feb  7 09:42:14 roothack kernel:        <ffffffff80110c23>{child_rip+8} <ffffffff8015f500>{kswapd+0}
Feb  7 09:42:14 roothack kernel:        <ffffffff80110c1b>{child_rip+0}
Feb  7 09:42:14 roothack kernel: journal_get_undo_access: No memory for committed data
Feb  7 09:42:14 roothack kernel: ext3_try_to_allocate_with_rsv: aborting transaction: Out of memory in __ext3_journal_get_undo_access
Feb  7 09:42:14 roothack kernel: EXT3-fs error (device sdd1) in ext3_new_block: Out of memory
Feb  7 09:42:14 roothack kernel: Aborting journal on device sdd1.
Feb  7 09:42:14 roothack kernel: ext3_abort called.
Feb  7 09:42:14 roothack kernel: EXT3-fs error (device sdd1): ext3_journal_start_sb: Detected aborted journal
Feb  7 09:42:14 roothack kernel: Remounting filesystem read-only
Feb  7 09:42:14 roothack kernel: EXT3-fs error (device sdd1) in ext3_ordered_writepage: Out of memory
Feb  7 09:42:14 roothack kernel: __journal_remove_journal_head: freeing b_committed_data
Feb  7 09:42:16 roothack kernel: EXT3-fs error (device sdd1) in start_transaction: Journal has aborted

This error may appear unrelated to VMware, but it seems that VMware can suck up all memory and starve EXT3 of enough memory to write the journal and memory can't be swapped to disk because that would cause more disk IO... Or something like that -- that doesn't seem quite right because SWAP partitions are different partitions and are not journaled.

Edit /etc/sysctl.conf and add this line

 vm.min_free_kbytes = 10240

Alternatively you can add this to /etc/rc.local or somewhere else in the boot process.

The value 10240 could be 5120 or higher.

For more information see the VMware forum.