Skip to content
Commits on Oct 21, 2010
  1. @seanlynch @steveyen

    Don't delete destination vbuckets before starting replication

    seanlynch committed with steveyen Oct 20, 2010
    This can cause deleted data to be resurrected if a node is
    failed. However, it avoids one path for memory overrun when starting
    replication.
    
    Change-Id: I91acd64d6c522303abbaf67ab4d17d2394a3d76a
    Reviewed-on: http://review.membase.org/3353
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
  2. @seanlynch @steveyen

    Don't delete destination vbuckets before transferring them

    seanlynch committed with steveyen Oct 20, 2010
    Deleting them and then dumping a lot of data into them can cause a
    memory overrun because persistence is currently blocked during
    deletion.
    
    This will result in deleted data's possibly being resurrected if the
    deletes occur during rebalance and a transfer is retried, or just
    before rebalance and a vbucket master is moved on top of a previous
    replica.
    
    Change-Id: I4353b37a02b2fba090c1485d241ef3bd6e08de10
    Reviewed-on: http://review.membase.org/3352
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
  3. @seanlynch @steveyen

    Only move one vbucket at a time.

    seanlynch committed with steveyen Oct 20, 2010
    This will minimize the impact of rebalance on performance.
    
    Change-Id: Ia71b6032b7aaf4cade7569e73f15d0fc0351a944
    Reviewed-on: http://review.membase.org/3351
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
  4. @seanlynch @steveyen

    Clean up vbucket deletion logic

    seanlynch committed with steveyen Oct 20, 2010
    We used to set to dead and delete on separate passes. There is no need
    to do that. I also made it easier to read.
    
    I'd thought there was a way for vbuckets to "leak" through in pending
    state, but it turns out they couldn't.
    
    Change-Id: I760cdcb8a8633e3e4f95afbcd5fc8940045fb535
    Reviewed-on: http://review.membase.org/3350
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
  5. @steveyen

    ratio is hits/gets not other way around. Bug 2798

    Aliaksey Kandratsenka committed with steveyen Oct 19, 2010
    Change-Id: Idad50af5ae0eb467e9c35a3dd848a7d0915db8ed
    Reviewed-on: http://review.membase.org/3348
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
Commits on Oct 13, 2010
  1. @seanlynch @dustin

    Use -a flag with stats commands

    seanlynch committed with dustin Oct 12, 2010
    Change-Id: Ifc6385f59ec4b5f999a4027de2d8b8b8a46a5e7d
    Reviewed-on: http://review.membase.org/3252
    Reviewed-by: Dustin Sallings <dustin@spy.net>
    Tested-by: Dustin Sallings <dustin@spy.net>
  2. @seanlynch @steveyen

    Remove trinary if-else for 2.4 compatibility

    seanlynch committed with steveyen Oct 12, 2010
    Change-Id: Ibce99b18c68ffd7e523a585a28f2b334b34da699
    Reviewed-on: http://review.membase.org/3245
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
  3. @seanlynch @steveyen

    Add memcached stats.

    seanlynch committed with steveyen Oct 12, 2010
    Change-Id: I40e8c7f971316ceb6bdcfbf93b52bd2f50aa2bb1
    Reviewed-on: http://review.membase.org/3243
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
Commits on Oct 12, 2010
  1. @steveyen @seanlynch

    bug 2763 - add collect_info to make bdist

    steveyen committed with seanlynch Oct 12, 2010
    Change-Id: I16a994f2db03b24565f1ed6a078653b5ff79e6d9
    Reviewed-on: http://review.membase.org/3237
    Reviewed-by: Sean Lynch <seanl@literati.org>
    Tested-by: Sean Lynch <seanl@literati.org>
Commits on Oct 11, 2010
  1. @steveyen

    update README northscale to membase

    steveyen committed Oct 11, 2010
    Change-Id: Idcbb45b4e3bc1ab5d5dbf98220ff918c6b4e4d01
    Reviewed-on: http://review.membase.org/3229
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
  2. @seanlynch @alk

    Python version of support data collector

    seanlynch committed with alk Oct 11, 2010
    This executes all the same commands as Trond's C version, but zips
    everything up and puts the browse_logs output in a separate log file.
    
    Change-Id: I53b9d2a7f6f3df6c0d681089a01719aecfb3093b
    Reviewed-on: http://review.northscale.com/3209
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  3. @seanlynch @alk

    Shut down memcached buckets when failing over a node

    seanlynch committed with alk Oct 11, 2010
    Change-Id: I646bf4d21cd62be791f54937582395082ffeec91
    Reviewed-on: http://review.northscale.com/3208
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  4. @steveyen

    fixed multi drawer behaviour

    Aliaksey Kandratsenka committed with steveyen Oct 11, 2010
    Change-Id: Ie6a8552cf4d78b08686e6be62020dc9daa54f5ac
    Reviewed-on: http://review.northscale.com/3206
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
  5. @seanlynch @steveyen

    Bug 2761: Don't EVER send the flush command.

    seanlynch committed with steveyen Oct 11, 2010
    This gets propagated over tap connections and deletes *all* data on
    any node that has replicas of vbuckets on the node where the command
    is sent.
    
    Change-Id: Iaab0947a7954f5e0a04d09a100d29ce28c554763
    Reviewed-on: http://review.northscale.com/3205
    Reviewed-by: Steve Yen <steve.yen@gmail.com>
    Tested-by: Steve Yen <steve.yen@gmail.com>
  6. @seanlynch

    tweaked texts of labels and messages

    Aliaksey Kandratsenka committed with seanlynch Oct 10, 2010
    Change-Id: I3d9123dc35b2707e584db3af0bdd72d79361234b
    Reviewed-on: http://review.northscale.com/3203
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
  7. @seanlynch

    don't display replica warnings for default bucket setup

    Aliaksey Kandratsenka committed with seanlynch Oct 10, 2010
    Previously when there were other validation errors we performed
    warnings validations event for default bucket setup call. This commit
    extracts parse_bucket_params_without_warnings function that default
    bucket setup parameters parsing function calls.
    
    Change-Id: I73a52ce4a5a31f60e6c09fa2bddfc27927c003f4
    Reviewed-on: http://review.northscale.com/3202
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
  8. @seanlynch

    decrease minimal membase bucket quota to 100 megs

    Aliaksey Kandratsenka committed with seanlynch Oct 10, 2010
    Change-Id: I55d93e0ed530f1c8c0ce3df914614caf92b46766
    Reviewed-on: http://review.northscale.com/3201
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
  9. @seanlynch @alk

    Add more user-visible logs to ns_memcached.

    seanlynch committed with alk Oct 10, 2010
    Change-Id: Iabb04ce6b3a44f2f0182485cbb2aedb6b0d53672
    Reviewed-on: http://review.northscale.com/3198
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  10. @seanlynch @alk

    Don't crash if a memcached bucket is already there.

    seanlynch committed with alk Oct 10, 2010
    Change-Id: If13d8c6c6e42cc37acebc3fb21b0b7a4e4b0bb4f
    Reviewed-on: http://review.northscale.com/3200
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  11. @seanlynch @alk

    Add some useful types to ns_common.hrl

    seanlynch committed with alk Oct 10, 2010
    Change-Id: I0d3f33eae56ce29fab846c445fd07d6faa75c696
    Reviewed-on: http://review.northscale.com/3199
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  12. @seanlynch

    UI change for bucket per node quota behaviour

    Aliaksey Kandratsenka committed with seanlynch Oct 10, 2010
    Change-Id: Ie6e0fd0a949433c6b310bff6160f2098c3a00c94
    Reviewed-on: http://review.northscale.com/3196
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
    Tested-by: Sean Lynch <seanl@literati.org>
  13. @seanlynch @alk

    Fix ns_bucket:live_bucket_nodes/1

    seanlynch committed with alk Oct 10, 2010
    It was returning lists of true.
    
    Change-Id: I1c0cc003c47154a7271433b06bc52f3caf325f17
    Reviewed-on: http://review.northscale.com/3193
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  14. @seanlynch

    run ns_bucket_sup:update_childs in another process. Bug 2737

    Aliaksey Kandratsenka committed with seanlynch Oct 10, 2010
    Diag for bug 2737 suggests that ns_bucket_sup has reached max restart
    frequency on some of ns_memcached childs and during it's resurrection
    by parent supervisor we're having ns_config_events doing update_childs
    call which in turn calls supervisor process at the same time as init
    function in that supervisor does call to ns_config_events making it a
    classic deadlock. The fix is to delegate update_childs to another
    process freeing ns_config_events for supervisor initialization.
    
    Change-Id: I85e27660f1f0352eada063b57d8b8b46287f58da
    Reviewed-on: http://review.northscale.com/3194
    Reviewed-by: Sean Lynch <seanl@literati.org>
    Tested-by: Sean Lynch <seanl@literati.org>
  15. @seanlynch @alk

    Delete bucket-0.sqlite also

    seanlynch committed with alk Oct 10, 2010
    Change-Id: Ifc70250138a08f77845e0990a460fbdbd98466ca
    Reviewed-on: http://review.northscale.com/3192
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  16. @seanlynch @alk

    Run ns_vbm_sup per bucket.

    seanlynch committed with alk Oct 10, 2010
    This causes vbucketmigrators to automatically shut down when a bucket
    is deleted.
    
    Change-Id: If1b2f18eda5ecc09bf1c8837ee3522825b74cd0b
    Reviewed-on: http://review.northscale.com/3180
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  17. @seanlynch @alk

    Bug 2736: Only run servers for buckets that are actually mapped

    seanlynch committed with alk Oct 10, 2010
    This will automatically shut down buckets when we fail a node. In
    addition, with the ns_vbm_sup change it will cause vbucketmigrators to
    shut down.
    
    Change-Id: If8eac8c54cff7dcb1f736a9161a0735f6fe7c432
    Reviewed-on: http://review.northscale.com/3179
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  18. @seanlynch

    fix thundering herd of requests from MultiDrawersWidget. Bug 2756

    Aliaksey Kandratsenka committed with seanlynch Oct 10, 2010
    We're avoiding creating new Cells that will fetch same bucket
    details. The price is slowly leaking that cells. More correct fix will
    be implemented later.
    
    Change-Id: I11d973c41952e5ab7363cc4574d5c1d690f768de
    Reviewed-on: http://review.northscale.com/3188
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
  19. @seanlynch

    fixed handling of janitor aborting

    Aliaksey Kandratsenka committed with seanlynch Oct 10, 2010
    Change-Id: I393dfa79c5a39b85683dfde31233804b85418869
    Reviewed-on: http://review.northscale.com/3186
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
  20. @seanlynch

    re-draw analytics after enabling updates. Bug 2726

    Aliaksey Kandratsenka committed with seanlynch Oct 8, 2010
    Change-Id: I8a34f109c3b8269d025df3e261ceb41ca7d9a63b
    Reviewed-on: http://review.northscale.com/3169
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
  21. @seanlynch @alk

    Only ask for stats from live nodes the bucket is mapped to.

    seanlynch committed with alk Oct 10, 2010
    Change-Id: I513276e01caafd4110f9630d860187fa1a285721
    Reviewed-on: http://review.northscale.com/3190
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  22. @seanlynch @alk

    Add ns_bucket:live_bucket_nodes/1

    seanlynch committed with alk Oct 10, 2010
    This will tell us which connected nodes a bucket is mapped to. Good
    for figuring out which nodes to ask for stats from.
    
    Change-Id: I9ae04b4da20dccea44e8980526fb3fa152babe71
    Reviewed-on: http://review.northscale.com/3189
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  23. @seanlynch @alk

    Flush buckets that are no longer mapped to this node as well.

    seanlynch committed with alk Oct 10, 2010
    Change-Id: Ia25c82a577aa907d30605e6d1e7cf6ef08bd5cee
    Reviewed-on: http://review.northscale.com/3187
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  24. @seanlynch @alk

    Wait for ns_memcached to be running to run the janitor

    seanlynch committed with alk Oct 10, 2010
    Change-Id: Ib479289fa84c6e054f14b0e2c724e1b69566e4de
    Reviewed-on: http://review.northscale.com/3182
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  25. @seanlynch @alk

    Bug 2736: Set servers to correct list of nodes before rebalancing

    seanlynch committed with alk Oct 10, 2010
    I've actually changed the order of operations so we only loop through
    the list of buckets once. This prevents us from loading all the
    buckets at once right at the beginning, which causes poor performance
    when there are a lot of buckets. The tradeoff is that now instead of a
    pause right at the beginning you'll see a pause every nth of the way
    through, where n is the number of buckets.
    
    Change-Id: Ida6877b13d07298256e7a782d9f76f0e1e16abc4
    Reviewed-on: http://review.northscale.com/3178
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
  26. @seanlynch @alk

    Move ns_mnesia:ensure_table into the gen_server

    seanlynch committed with alk Oct 10, 2010
    We were having crashes from trying to create tables while shutting
    down.
    
    Change-Id: I5dc2f951abc6036558d040890e821ed34a838594
    Reviewed-on: http://review.northscale.com/3184
    Reviewed-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
    Tested-by: Aliaksey Kandratsenka <alkondratenko@gmail.com>
Something went wrong with that request. Please try again.