Re: RBD on ec pool with compression.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



This is "supposed" to work, but the compression in Bluestore has less testing than most things there and is pretty invasive, so when I discussed this with Radoslaw (added) last week there were some obvious places to look. Hopefully it's not too hard to identify the problem from these backtraces and get a fix in? :)
-Greg

On Fri, Oct 20, 2017 at 12:16 AM Cassiano Pilipavicius <cassiano@xxxxxxxxxxx> wrote:
Hello, is it possible to use compression on a EC pool? I am trying to
enable this to use as a huge backup/archive disk, the data is almost
static and access to it is very sporadic, so, bad performance is not a
concern here.

I've created the RBD storing data to the EC pool (--data-pool option)
and enabled compression on the EC pool, but as soon as I start to write
data to the volume, I have a crash on 12 osds (my ec ruleset is k=9 m=3)

If I use only RBD+EC it works fine, if I use RBD+Compression (no erasure
coding) it also works fine.

This is what appears on the log of the crashed osds:


  ceph version 12.2.1 (3e7492b9ada8bdc9a5cd0feafd42fbca27f9c38e)
luminous (stable)
  1: (ceph::__ceph_assert_fail(char const*, char const*, int, char
const*)+0x110) [0x7f019377af20]
  2: (bluestore_blob_t::map(unsigned long, unsigned long,
std::function<int (unsigned long, unsigned long)>) const+0xfe)
[0x7f019365f14e]
  3: (BlueStore::_do_remove(BlueStore::TransContext*,
boost::intrusive_ptr<BlueStore::Collection>&,
boost::intrusive_ptr<BlueStore::Onode>)+0x1151) [0x7f019363d191]
  4: (BlueStore::_remove(BlueStore::TransContext*,
boost::intrusive_ptr<BlueStore::Collection>&,
boost::intrusive_ptr<BlueStore::Onode>&)+0x94) [0x7f019363da64]
  5: (BlueStore::_txc_add_transaction(BlueStore::TransContext*,
ObjectStore::Transaction*)+0x15af) [0x7f019365047f]
  6: (BlueStore::queue_transactions(ObjectStore::Sequencer*,
std::vector<ObjectStore::Transaction,
std::allocator<ObjectStore::Transaction> >&,
boost::intrusive_ptr<TrackedOp>, ThreadPool::TPHandle*)+0x3a0)
[0x7f01936513a0]
  7:
(PrimaryLogPG::queue_transactions(std::vector<ObjectStore::Transaction,
std::allocator<ObjectStore::Transaction> >&,
boost::intrusive_ptr<OpRequest>)+0x65) [0x7f01933c1f35]
  8: (ECBackend::handle_sub_write(pg_shard_t,
boost::intrusive_ptr<OpRequest>, ECSubWrite&, ZTracer::Trace const&,
Context*)+0x631) [0x7f01934e1cc1]
  9: (ECBackend::_handle_message(boost::intrusive_ptr<OpRequest>)+0x327)
[0x7f01934f2867]
  10: (PGBackend::handle_message(boost::intrusive_ptr<OpRequest>)+0x50)
[0x7f01933f73d0]
  11: (PrimaryLogPG::do_request(boost::intrusive_ptr<OpRequest>&,
ThreadPool::TPHandle&)+0x5ae) [0x7f0193362fbe]
  12: (OSD::dequeue_op(boost::intrusive_ptr<PG>,
boost::intrusive_ptr<OpRequest>, ThreadPool::TPHandle&)+0x3f9)
[0x7f01931f33a9]
  13: (PGQueueable::RunVis::operator()(boost::intrusive_ptr<OpRequest>
const&)+0x57) [0x7f0193465797]
  14: (OSD::ShardedOpWQ::_process(unsigned int,
ceph::heartbeat_handle_d*)+0xfce) [0x7f019321e9ee]
  15: (ShardedThreadPool::shardedthreadpool_worker(unsigned int)+0x839)
[0x7f0193780a39]
  16: (ShardedThreadPool::WorkThreadSharded::entry()+0x10) [0x7f01937829d0]
  17: (()+0x7dc5) [0x7f019013adc5]
  18: (clone()+0x6d) [0x7f018f22e28d]
  NOTE: a copy of the executable, or `objdump -rdS <executable>` is
needed to interpret this.

--- logging levels ---
    0/ 5 none
    0/ 0 lockdep
    0/ 0 context
    0/ 0 crush
    1/ 5 mds
    1/ 5 mds_balancer
    1/ 5 mds_locker
    1/ 5 mds_log
    1/ 5 mds_log_expire
    1/ 5 mds_migrator
    0/ 0 buffer
    0/ 0 timer
    0/ 1 filer
    0/ 1 striper
    0/ 1 objecter
    0/ 5 rados
    0/ 5 rbd
    0/ 5 rbd_mirror
    0/ 5 rbd_replay
    0/ 0 journaler
    0/ 5 objectcacher
    0/ 5 client
    0/ 0 osd
    0/ 0 optracker
    0/ 0 objclass
    0/ 0 filestore
    0/ 0 journal
    0/ 0 ms
    1/ 5 mon
    0/ 0 monc
    1/ 5 paxos
    0/ 0 tp
    0/ 0 auth
    1/ 5 crypto
    0/ 0 finisher
    0/ 0 heartbeatmap
    0/ 0 perfcounter
    1/ 5 rgw
    1/10 civetweb
    1/ 5 javaclient
    0/ 0 asok
    0/ 0 throttle
    0/ 0 refs
    1/ 5 xio
    1/ 5 compressor
    1/ 5 bluestore
    1/ 5 bluefs
    1/ 3 bdev
    1/ 5 kstore
    4/ 5 rocksdb
    4/ 5 leveldb
    4/ 5 memdb
    1/ 5 kinetic
    1/ 5 fuse
    1/ 5 mgr
    1/ 5 mgrc
    1/ 5 dpdk
    1/ 5 eventtrace
   -2/-2 (syslog threshold)
   -1/-1 (stderr threshold)
   max_recent     10000
   max_new         1000
   log_file /var/log/ceph/ceph-osd.8.log
--- end dump of recent events ---
2017-10-20 04:28:51.564373 7f0176684700 -1 *** Caught signal (Aborted) **
  in thread 7f0176684700 thread_name:tp_osd_tp

  ceph version 12.2.1 (3e7492b9ada8bdc9a5cd0feafd42fbca27f9c38e)
luminous (stable)
  1: (()+0xa29511) [0x7f019373c511]
  2: (()+0xf100) [0x7f0190142100]
  3: (gsignal()+0x37) [0x7f018f16d5f7]
  4: (abort()+0x148) [0x7f018f16ece8]
  5: (ceph::__ceph_assert_fail(char const*, char const*, int, char
const*)+0x284) [0x7f019377b094]
  6: (bluestore_blob_t::map(unsigned long, unsigned long,
std::function<int (unsigned long, unsigned long)>) const+0xfe)
[0x7f019365f14e]
  7: (BlueStore::_do_remove(BlueStore::TransContext*,
boost::intrusive_ptr<BlueStore::Collection>&,
boost::intrusive_ptr<BlueStore::Onode>)+0x1151) [0x7f019363d191]
  8: (BlueStore::_remove(BlueStore::TransContext*,
boost::intrusive_ptr<BlueStore::Collection>&,
boost::intrusive_ptr<BlueStore::Onode>&)+0x94) [0x7f019363da64]
  9: (BlueStore::_txc_add_transaction(BlueStore::TransContext*,
ObjectStore::Transaction*)+0x15af) [0x7f019365047f]
  10: (BlueStore::queue_transactions(ObjectStore::Sequencer*,
std::vector<ObjectStore::Transaction,
std::allocator<ObjectStore::Transaction> >&,
boost::intrusive_ptr<TrackedOp>, ThreadPool::TPHandle*)+0x3a0)
[0x7f01936513a0]
  11:
(PrimaryLogPG::queue_transactions(std::vector<ObjectStore::Transaction,
std::allocator<ObjectStore::Transaction> >&,
boost::intrusive_ptr<OpRequest>)+0x65) [0x7f01933c1f35]
  12: (ECBackend::handle_sub_write(pg_shard_t,
boost::intrusive_ptr<OpRequest>, ECSubWrite&, ZTracer::Trace const&,
Context*)+0x631) [0x7f01934e1cc1]
  13:
(ECBackend::_handle_message(boost::intrusive_ptr<OpRequest>)+0x327)
[0x7f01934f2867]
  14: (PGBackend::handle_message(boost::intrusive_ptr<OpRequest>)+0x50)
[0x7f01933f73d0]
  15: (PrimaryLogPG::do_request(boost::intrusive_ptr<OpRequest>&,
ThreadPool::TPHandle&)+0x5ae) [0x7f0193362fbe]
  16: (OSD::dequeue_op(boost::intrusive_ptr<PG>,
boost::intrusive_ptr<OpRequest>, ThreadPool::TPHandle&)+0x3f9)
[0x7f01931f33a9]
  17: (PGQueueable::RunVis::operator()(boost::intrusive_ptr<OpRequest>
const&)+0x57) [0x7f0193465797]
  18: (OSD::ShardedOpWQ::_process(unsigned int,
ceph::heartbeat_handle_d*)+0xfce) [0x7f019321e9ee]
  19: (ShardedThreadPool::shardedthreadpool_worker(unsigned int)+0x839)
[0x7f0193780a39]
  20: (ShardedThreadPool::WorkThreadSharded::entry()+0x10) [0x7f01937829d0]
  21: (()+0x7dc5) [0x7f019013adc5]
  22: (clone()+0x6d) [0x7f018f22e28d]
  NOTE: a copy of the executable, or `objdump -rdS <executable>` is
needed to interpret this.

--- begin dump of recent events ---
      0> 2017-10-20 04:28:51.564373 7f0176684700 -1 *** Caught signal
(Aborted) **
  in thread 7f0176684700 thread_name:tp_osd_tp

  ceph version 12.2.1 (3e7492b9ada8bdc9a5cd0feafd42fbca27f9c38e)
luminous (stable)
  1: (()+0xa29511) [0x7f019373c511]
  2: (()+0xf100) [0x7f0190142100]
  3: (gsignal()+0x37) [0x7f018f16d5f7]
  4: (abort()+0x148) [0x7f018f16ece8]
  5: (ceph::__ceph_assert_fail(char const*, char const*, int, char
const*)+0x284) [0x7f019377b094]
  6: (bluestore_blob_t::map(unsigned long, unsigned long,
std::function<int (unsigned long, unsigned long)>) const+0xfe)
[0x7f019365f14e]
  7: (BlueStore::_do_remove(BlueStore::TransContext*,
boost::intrusive_ptr<BlueStore::Collection>&,
boost::intrusive_ptr<BlueStore::Onode>)+0x1151) [0x7f019363d191]
  8: (BlueStore::_remove(BlueStore::TransContext*,
boost::intrusive_ptr<BlueStore::Collection>&,
boost::intrusive_ptr<BlueStore::Onode>&)+0x94) [0x7f019363da64]
  9: (BlueStore::_txc_add_transaction(BlueStore::TransContext*,
ObjectStore::Transaction*)+0x15af) [0x7f019365047f]
  10: (BlueStore::queue_transactions(ObjectStore::Sequencer*,
std::vector<ObjectStore::Transaction,
std::allocator<ObjectStore::Transaction> >&,
boost::intrusive_ptr<TrackedOp>, ThreadPool::TPHandle*)+0x3a0)
[0x7f01936513a0]
  11:
(PrimaryLogPG::queue_transactions(std::vector<ObjectStore::Transaction,
std::allocator<ObjectStore::Transaction> >&,
boost::intrusive_ptr<OpRequest>)+0x65) [0x7f01933c1f35]
  12: (ECBackend::handle_sub_write(pg_shard_t,
boost::intrusive_ptr<OpRequest>, ECSubWrite&, ZTracer::Trace const&,
Context*)+0x631) [0x7f01934e1cc1]
  13:
(ECBackend::_handle_message(boost::intrusive_ptr<OpRequest>)+0x327)
[0x7f01934f2867]
  14: (PGBackend::handle_message(boost::intrusive_ptr<OpRequest>)+0x50)
[0x7f01933f73d0]
  15: (PrimaryLogPG::do_request(boost::intrusive_ptr<OpRequest>&,
ThreadPool::TPHandle&)+0x5ae) [0x7f0193362fbe]
  16: (OSD::dequeue_op(boost::intrusive_ptr<PG>,
boost::intrusive_ptr<OpRequest>, ThreadPool::TPHandle&)+0x3f9)
[0x7f01931f33a9]
  17: (PGQueueable::RunVis::operator()(boost::intrusive_ptr<OpRequest>
const&)+0x57) [0x7f0193465797]
  18: (OSD::ShardedOpWQ::_process(unsigned int,
ceph::heartbeat_handle_d*)+0xfce) [0x7f019321e9ee]
  19: (ShardedThreadPool::shardedthreadpool_worker(unsigned int)+0x839)
[0x7f0193780a39]
  20: (ShardedThreadPool::WorkThreadSharded::entry()+0x10) [0x7f01937829d0]
  21: (()+0x7dc5) [0x7f019013adc5]
  22: (clone()+0x6d) [0x7f018f22e28d]
  NOTE: a copy of the executable, or `objdump -rdS <executable>` is
needed to interpret this.

--- logging levels ---
    0/ 5 none
    0/ 0 lockdep
    0/ 0 context
    0/ 0 crush
    1/ 5 mds
    1/ 5 mds_balancer
    1/ 5 mds_locker
    1/ 5 mds_log
    1/ 5 mds_log_expire
    1/ 5 mds_migrator
    0/ 0 buffer
    0/ 0 timer
    0/ 1 filer
    0/ 1 striper
    0/ 1 objecter
    0/ 5 rados
    0/ 5 rbd
    0/ 5 rbd_mirror
    0/ 5 rbd_replay
    0/ 0 journaler
    0/ 5 objectcacher
    0/ 5 client
    0/ 0 osd
    0/ 0 optracker
    0/ 0 objclass
    0/ 0 filestore
    0/ 0 journal
    0/ 0 ms
    1/ 5 mon
    0/ 0 monc
    1/ 5 paxos
    0/ 0 tp
    0/ 0 auth
    1/ 5 crypto
    0/ 0 finisher
    0/ 0 heartbeatmap
    0/ 0 perfcounter
    1/ 5 rgw
    1/10 civetweb
    1/ 5 javaclient
    0/ 0 asok
    0/ 0 throttle
    0/ 0 refs
    1/ 5 xio
    1/ 5 compressor
    1/ 5 bluestore
    1/ 5 bluefs
    1/ 3 bdev
    1/ 5 kstore
    4/ 5 rocksdb
    4/ 5 leveldb
    4/ 5 memdb
    1/ 5 kinetic
    1/ 5 fuse
    1/ 5 mgr
    1/ 5 mgrc
    1/ 5 dpdk
    1/ 5 eventtrace
   -2/-2 (syslog threshold)
   -1/-1 (stderr threshold)
   max_recent     10000
   max_new         1000
   log_file /var/log/ceph/ceph-osd.8.log
--- end dump of recent events ---

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

[Index of Archives]     [Information on CEPH]     [Linux Filesystem Development]     [Ceph Development]     [Ceph Large]     [Linux USB Development]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [xfs]


  Powered by Linux