1. 24 Jun, 2016 1 commit
    • Kirill Smelkov's avatar
      bigfile/pagemap: Fix non-leaf page iteration · ee9bcd00
      Kirill Smelkov authored
      Since the beginning of pagemap (45af76e6 "bigfile/pagemap: specialized
      {} uint64 -> void * mapping") we had a bug sitting in
      __pagemap_for_each_leaftab() (non-leaf iterating logic behind
      pagemap_for_each):
      
      After entry to stack-down was found, we did not updated tailv[l]
      accordingly. Thus if there are non-adjacent entries an entry could be
      e.g. emitted many times:
      
           l 3  __down 0x7f79da1ee000
           tailv[4]: 0x7f79da1ee000
            -> tailv[4] 0x7f79da1ee000  __down 0x7f79da1ed000
      
           l 4  __down 0x7f79da1ed000
           tailv[5]: 0x7f79da1ed000
           h 5  l 5  leaftab: 0x7f79da1ed000      <--
            lvl 5  idx 169  page 0x55aa
          ok 9 - pagemap_for_each(0) == 21930
      
           l 5  __down (nil)
           tailv[4]: 0x7f79da1ee008
            -> tailv[4] 0x7f79da1ee008  __down 0x7f79da1ed000
      
           l 4  __down 0x7f79da1ed000
           tailv[5]: 0x7f79da1ed000
           h 5  l 5  leaftab: 0x7f79da1ed000      <--
            lvl 5  idx 169  page 0x55aa
          not ok 10 - pagemap_for_each(1) == 140724106500272
      
      And many-time-emitted entries are not only incorrect, but can also lead
      to not-handled segmentation faults in e.g. fileh_close():
      
          https://lab.nexedi.com/nexedi/wendelin.core/blob/v0.6-1-gb0b2c52/bigfile/virtmem.c#L179
      
          /* drop all pages (dirty or not) associated with this fileh */
          pagemap_for_each(page, &fileh->pagemap) {
              /* it's an error to close fileh to mapping of which an access is
               * currently being done in another thread */
              BUG_ON(page->state == PAGE_LOADING);
              page_drop_memory(page);
              list_del(&page->lru);                           <-- HERE
              bzero(page, sizeof(*page)); /* just in case */
              free(page);
          }
      
      ( because after first bzero of a page, the page is all 0 bytes including
        page->lru{.next,.prev} so on the second time when the same page is
        emitted by pagemap_for_each, list_del(&page->lru) will try to set
        page->lru.next = ... which will segfault. )
      
      So fix it by properly updating tailv[l] while we scan/iterate current level.
      
      NOTE
      
      This applies only to non-leaf pagemap levels, as leaf level is scanned
      with separate loop in pagemap_for_each. That's why we probably did not
      noticed this earlier - up until now our usual workloads was to change
      data in adjacent batches and that means adjacent pages.
      
      Though today @Tyagov was playing with wendelin.core in some other way and
      it uncovered the bug.
      ee9bcd00
  2. 13 Jun, 2016 5 commits
  3. 12 Jun, 2016 1 commit
    • Kirill Smelkov's avatar
      3rdparty/ccan: Update · 89d099af
      Kirill Smelkov authored
      Just update to latest CCAN for it to be a fresh one.
      
      Throught the modules we use there are no real updates, just a fix for
      one warning in array_size.
      89d099af
  4. 20 Apr, 2016 1 commit
    • Kirill Smelkov's avatar
      bigfile/zodb: Don't write ZBlk to DB if it was not changed · 6773fda0
      Kirill Smelkov authored
      For ZBlk1 we already compare ZData content about whether it was changed
      compared to data already stored to DB, and do not store it twice if data
      is the same.
      
      However ZBlk itself is always marked as changed, if corresponding memory
      page was dirtied. This results in transactions like
      
      Trans #33915309 tid=03b6944919befeee time=2016-04-17 22:01:06.034237 offset=140320105842
          status=' ' user='...' description='...'
        # ... other parts, but no ZData here
        data #00002 oid=000000000026fc4c size=79 class=wendelin.bigfile.file_zodb.ZBlk1
      
      where ZBlk1 is committed the same without necessity.
      
      NOTE we cannot avoid committing ZBlk in all cases, because it is used to signal
          other DB clients that a ZBlk needs to be invalidated and this way associated
          fileh pages are invalidated too.
      
          This cannot work via ZData, because ZData don't have back-pointer to
          ZBlk1 or to corresponding zfile.
      6773fda0
  5. 05 Apr, 2016 1 commit
    • Kirill Smelkov's avatar
      Fix develop install for setuptools >= 19.4 · 2ce96a76
      Kirill Smelkov authored
      Starting from setuptools 19.4, more concrete from the following commit:
      
          https://github.com/pypa/setuptools/commit/ebc54982
      
      setuptools sorts namespaced packages .__path__ to be in sync with sys.path .
      That however breaks for wendelin.core used from in-tree or installed in
      development mode, because we are doing tricks in top-level import redirector
      (see e870781d "Top-level in-tree import redirector"):
      
          (z+numpy.v2)kirr@teco:~/tmp/trashme/wendelin.core$ python -c 'import wendelin'
          Traceback (most recent call last):
            File "<string>", line 1, in <module>
            File "wendelin.py", line 39, in <module>
              __import__('pkg_resources').declare_namespace(__name__)
            File "/home/kirr/src/wendelin/venv/z+numpy.v2/local/lib/python2.7/site-packages/pkg_resources/__init__.py", line 2081, in declare_namespace
              _handle_ns(packageName, path_item)
            File "/home/kirr/src/wendelin/venv/z+numpy.v2/local/lib/python2.7/site-packages/pkg_resources/__init__.py", line 2026, in _handle_ns
              _rebuild_mod_path(path, packageName, module)
            File "/home/kirr/src/wendelin/venv/z+numpy.v2/local/lib/python2.7/site-packages/pkg_resources/__init__.py", line 2050, in _rebuild_mod_path
              orig_path.sort(key=position_in_sys_path)
            File "/home/kirr/src/wendelin/venv/z+numpy.v2/local/lib/python2.7/site-packages/pkg_resources/__init__.py", line 2045, in position_in_sys_path
              return sys_path.index(_normalize_cached(os.sep.join(parts)))
          ValueError: '/home/kirr/tmp/trashme' is not in list
      
      Here wendelin.py added /home/kirr/tmp/trashme/wendelin.core to .__path__ and
      setuptools' _handle_ns() wants to order that dir's parent in correspondence
      with sys.path, but parent path is not there - oops.
      
      We can workaround the problem, by first not initializing .__path__ and letting
      
          __import__('pkg_resources').declare_namespace(__name__)
      
      fully handle and initialize it, and only after it is done we make the
      correction for wendelin modules located not under .../wendelin.core/wendelin/
      but under .../wendelin.core/ .
      
      Importing was tested to work with the fix with both setuptools 20.6.7 and older
      setuptools 17.1.1, i.e. here we should not be breaking backward compatibility.
      
      /reported-by @tatuya, @Camata, @Tyagov
      
      /reviewed-on !1
      2ce96a76
  6. 18 Dec, 2015 1 commit
    • Kirill Smelkov's avatar
      ZBigArray: Compatibility fix to read arrays from DB that were previously saved without order info · 2ca0f076
      Kirill Smelkov authored
      Commit ab9ca2df (bigarray: Add support for FORTRAN ordering) added
      ability to define array order, but there I made a mistake of not caring
      about how previously-saved to DB arrays would be read back.
      
      The thing is BigArray gained new data member ._order which is
      automatically saved to DB thanks to ZBigArray inheriting from
      Persistent; on load-from-db path we just read object state from DB,
      which for ZBigArray is dict, and restore object attributes from it.
      
      But for previously-saved data, obviously, there is no 'order' entry and thus
      this way restored objects are restored not in full to current code expectations
      and it can boom e.g. this way:
      
          zarray.resize((new_one,old_shape[1]))
        Module wendelin.bigarray, line 190, in resize
          self._init0(new_shape, self.dtype, order=self._order)
        AttributeError: 'ZBigArray' object has no attribute '_order'
      
      Solution to fix is: on restore-from-DB path, see if a data member is not
      present on restored object, and if it has default value in BigArray set it to
      that.
      
      ( code to get function defaults is from
        http://stackoverflow.com/questions/12627118/get-a-function-arguments-default-value )
      
      /cc @Tyagov, @klaus
      2ca0f076
  7. 15 Dec, 2015 6 commits
    • Kirill Smelkov's avatar
      tox: Bump NEO to 1.6 · 18b40b18
      Kirill Smelkov authored
      18b40b18
    • Kirill Smelkov's avatar
      bigfile/virtmem: Do loadblk() with virtmem lock released · f49c11a3
      Kirill Smelkov authored
      loadblk() calls are potentially slow and external code that serve the cal can
      take other locks in addition to virtmem lock taken by virtmem subsystem. If
      that "other locks" are also taken before external code calls e.g.
      fileh_invalidate_page() in different codepath a deadlock can happen, e.g.
      
            T1                  T2
      
            page-access         invalidation-from-server received
            V -> loadblk
                                Z   <- ClientStorage.invalidateTransaction()
            Z -> zeo.load
                                V   <- fileh_invalidate_page
      
      The solution to avoid deadlock is to call loadblk() with virtmem lock released
      and upon loadblk() completion recheck virtmem data structures carefully.
      
      To make that happen:
      
      - new page state is introduces:
      
          PAGE_LOADING                (file content loading is  in progress)
      
      - virtmem releases virt_lock before calling loadblk() when serving pagefault
      
      - because loading is now done with virtmem lock released, now:
      
      1. After loading completes we need to recheck fileh/vma data structures
      
         The recheck is done in full - vma_on_pagefault() just asks its driver (see
         VM_RETRY and VM_HANDLED codes) to retry handling the fault completely. This
         should work as the freshly loaded page was just inserted into fileh->pagemap
         and should be found there in the cache on next lookup.
      
         On the other hand this also works correctly, if there was concurrent change
         - e.g. vma was unmapped while we were loading the data - in that case the
         fault will be also processed correctly - but loaded data will stay in
         fileh->pagemap (and if not used will be evicted as not-needed
         eventually by RAM reclaim).
      
      2. Similar to retrying mechanism is used for cases when two threads
         concurrently access the same page and would both try to load corresponding
         block - only one thread issues the actual loadblk() and another waits for load
         to complete with polling and VM_RETRY.
      
      3. To correctly invalidate loading-in-progress pages another new page state
         is introduced:
      
          PAGE_LOADING_INVALIDATED    (file content loading was in progress
                                       while request to invalidate the page came in)
      
         which fileh_invalidate_page() uses to propagate invalidation message to
         loadblk() caller.
      
      4. Blocks loading can now happen in parallel with other block loading and
         other virtmem operations - e.g. invalidation. For such cases tests are added
         to test_thread.py
      
      5. virtmem lock now becomes just regular lock, instead of being previously
         recursive.
      
         For virtmem lock to be recursive was needed for cases, when code under
         loadblk() could trigger other virtmem calls, e.g. due to GC and calling
         another VMA dtor that would want to lock virtmem, but virtmem lock was
         already held.
      
         This is no longer needed.
      
      6. To catch double faults we now cannot use just on static variable
         in_on_pagefault. That variable thus becomes thread-local.
      
      7. Old test in test_thread to "test that access vs access don't overlap" no
         longer holds true - and is thus removed.
      
      /cc @Tyagov, @klaus
      f49c11a3
    • Kirill Smelkov's avatar
      bigfile/virtmem: Factor functionality to unlock/retake GIL into own functions · 0231a65d
      Kirill Smelkov authored
      Previously we were doing virt_lock() / virt_unlock() which automatically
      were making sure to unlock GIL before locking virtmem, and to restore
      GIL state to previous after virtmem lock happened. virt_unlock() was
      unlocking just the virtmem lock without touching GIL at all - that works
      because the running code would eventually release GIL as python
      regularly does so to allowing multiple threads to run.
      
      In the next patch however, we'll need to wait for in-progress-loading
      page to complete, and that wait has to be done with GIL released (so
      other python threads could run), and for doing so we'll need
      functionality to make sure GIL is unlocked and retake it back, not tied
      to virt_lock().
      
      So factor it out.
      0231a65d
    • Kirill Smelkov's avatar
      bigfile/tests: move NotifyChannel to test_thread.py · 99cd1f03
      Kirill Smelkov authored
      NotifyChannel was introduced in c7c01ce4 (bigfile/zodb: ZODB.Connection
      can migrate between threads on close/open and we have to care) to test
      thread interaction specific to ZODB.
      
      We'll however need NotifyChannel to do more threading test of virtmem
      core, and this way the proper place for NotifyChannel is test_thread.py
      itself.
      
      Move it.
      99cd1f03
    • Kirill Smelkov's avatar
      bigfile/virtmem: Remove obsolete XXX about locking · 81bf620c
      Kirill Smelkov authored
      Both comments are from the beginning - from 9a293c2d (bigfile/virtmem:
      Userspace Virtual Memory Manager) - but d53271b9 patch (bigfile/virtmem:
      Big Virtmem lock) missed to update them.
      81bf620c
    • Kirill Smelkov's avatar
      bigfile: Plug memory leak in ramh_close() · 997ebacd
      Kirill Smelkov authored
      No one was freeing RAMH structure itself, and thus ASAN reports e.g.:
      
      ==15935==ERROR: LeakSanitizer: detected memory leaks
      
      Direct leak of 32 byte(s) in 1 object(s) allocated from:
          #0 0x7f29c89f1001 in __interceptor_calloc (/usr/lib/x86_64-linux-gnu/libasan.so.2+0x94001)
          #1 0x401da1 in zalloc include/wendelin/utils.h:65
          #2 0x408128 in shmfs_ramh_open bigfile/tests/../ram_shmfs.c:202
          #3 0x407611 in ramh_open bigfile/tests/../ram.c:81
          #4 0x402560 in fileh_open bigfile/tests/../virtmem.c:131
          #5 0x427ca1 in test_pagefault_savestate bigfile/tests/test_virtmem.c:1022
          #6 0x4281ba in main bigfile/tests/test_virtmem.c:1061
          #7 0x7f29c83b8b44 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x21b44)
      
      NOTE similar leak remains open in ram_close(), but it is a bit involved
      to fix and the effort will be removed anyway after we switch to kernel
      virtual memory manager. Besides ramh are opened and closed all the time
      and ram only once.
      997ebacd
  8. 05 Nov, 2015 1 commit
    • Kirill Smelkov's avatar
      bigfile/ram_shmfs: Warn users explicitly if fallocate() is not supported · e8c05a22
      Kirill Smelkov authored
      Currently wendelin.core does not work on e.g. Debian 7, because that
      distro has too old kernel without support for fallocate on tmpfs.
      But the diagnostics of failure is not clear and looks like just being
      out of memory:
      
          bigfile/tests/../virtmem.c:845 OOM      BUG!
      
      what happens in fact is that
      
          - virtmem tries to allocate a page -> calls shmfs_alloc_page(),
          - fallocate() fails with "operation not supported" error code
          - virtmem sees this as page allocation failure,
          - tries to reclaim pages,
          - but there are no allocated pages at all -> OOM
      
      Detect whether fallocate() error is operational error, or simply
      "fallocate not supported" and if latter, report to user. Now it looks
      like:
      
          bigfile/tests/../ram_shmfs.c:129 shmfs_alloc_page WARN: fallocate() not supported
          bigfile/tests/../virtmem.c:845 OOM      BUG!
      
      /cc @Tyagov
      e8c05a22
  9. 02 Nov, 2015 1 commit
  10. 02 Oct, 2015 9 commits
  11. 30 Sep, 2015 1 commit
    • Kirill Smelkov's avatar
      bigfile/zodb: Teach ZBlk1 not to waste memory on memory-page -> DB path · 1bf0cf31
      Kirill Smelkov authored
      ZBlk* objects are intermediate ZODB object in between data stored in
      ZODB and memory pages managed by virtmem. As such, after they do their
      job to either load data from DB to memory, or store from memory to DB,
      it is not needed to keep them alive with duplicate content thus only
      wasting memory.
      
      ZBlk0 cares about this detail via "deactivating" ._v_blkdata in
      loadblkdata() and __getstate__() prologues.
      
      ZBlk1 did the same for load path in loadblkdata() prologue, but for
      .__getstate__() it was not directly possible, because for ZBlk1 the
      state is IOBTree, not one non-persistent object, and thus it first needs
      to be processed by ZODB together with its subobjects on its way to
      storage and only then all they deactivated.
      
      So 13c0c17c (bigfile/zodb: Format #1 which is optimized for small
      changes) only put TODO for memory-page -> DB path about not wasting
      memory this way.
      
      But the problem is relatively easy to solve:
      
          - we can deactivate ZData objects (leaf objects in ZBlk1.chunktab
            btree) by hooking into ZData.__getstate__() prologue;
      
          - we also need to care to deactivate chunks right away, which
            setblkdata() loaded to compare .data and found them to be not
            changed
      
      This way we do not waste memory keeping intermediate ZData objects alive
      with the same content as memory page after commit.
      
      /cc @Tyagov
      1bf0cf31
  12. 28 Sep, 2015 5 commits
  13. 24 Sep, 2015 7 commits
    • Kirill Smelkov's avatar
      bigfile/zodb: Format #1 which is optimized for small changes · 13c0c17c
      Kirill Smelkov authored
      Our current approach is that each file block is represented by 1 zodb
      object, with block size being 2M. Even with trailing \0 trimming, which
      halves the overhead on average, DB size grows very fast if we do a lot
      of small appends or changes. So another format needs to be introduced
      which has lower overhead for storing small changes:
      
      In general, to represent BigFile as ZODB objects, each file block could
      be represented separately either as
      
          1) one ZODB object, or          (ZBlk0 - this what we have already)
          2) group of ZODB objects        (ZBlk1 - this is what we introduce)
      
      with top-level BTree directory #blk -> objects representing block.
      
      For "1" we have
      
          - low-overhead access time (only 1 object loaded from DB), but
          - high-overhead in terms of ZODB size (with FileStorage / ZEO, every change
            to a block causes it to be written into DB in full again)
      
      For "2" we have
      
          - low-overhead in terms of ZODB size (only part of a block is overwritten
            in DB on single change), but
          - high-overhead in terms of access time
            (several objects need to be loaded for 1 block)
      
      In general it is not possible to have low-overhead for both i) access-time, and
      ii) DB size, with approach where we do block objects representation /
      management on *client* side.
      
      On the other hand, if object management is moved to DB *server* side, it is
      possible to deduplicate them there and this way have low-overhead for both
      access-time and DB size with just client storing 1 object per file block. This
      will be our future approach after we teach NEO about object deduplication.
      
      ~~~~
      
      As shown above in the last paragraph it is not possible to perform
      optimally on client side. Thus ZBlk1 should be only an intermediate
      solution until we move data management to DB server side, with main
      criteria for ZBlk1 to keep it simple.
      
      In this patch a simple scheme is used, where every block is divided into
      chunks organized via BTree. When a block part changes, only corresponding
      chunk is updated. Chunk size is chosen to be 4K which creates ~ 512
      fanout for 2M block.
      
      DB size after tests is changed as follows:
      
              bigfile     bigarray
      
      ZBlk0     24K       6200K
      ZBlk1     36K         36K
      
      ( slight size increase for bigfile tests is because of btree structures
        overhead )
      
      Time to run tests stays approximately the same.
      
      /cc @Tyagov, @klaus
      13c0c17c
    • Kirill Smelkov's avatar
      bigfile/zodb: Prepare to have several ZBlk formats · 70ea8573
      Kirill Smelkov authored
      - current ZBlk becomes format 0
      - write format can be selected via WENDELIN_CORE_ZBLK_FMT env var
      - upon writing a block we always make sure we write it in current write
        format - so if a block was previously written in one format, it could
        be changed on the next write.
      - tox is prepared to test all write formats (so far only ZBlk0 there).
      
      The reason is - in the next patch we'll introduce another format for
      blocks which is optimized for small changes.
      70ea8573
    • Kirill Smelkov's avatar
      bigfile/zodb: Split common functionality from ZBlk to base class · a7b7c294
      Kirill Smelkov authored
      If we aim to have several kinds of ZBlk, the functionality to invalidate
      a block and bind it to zfile is common and thus should be shared.
      
      If we introduce a base class, it also makes sense to document what
      .loadblkdata() and .setblkdata() should do there - in one place.
      a7b7c294
    • Kirill Smelkov's avatar
      bigfile/zodb: Reuse __setstate__() in ZBlk.__init__() · a354d36b
      Kirill Smelkov authored
      - we have logic to init ._v_zfile and ._v_blk there
      
      - the same for ._v_blkdata - logic to init it is there + it is better to
        set variables right from instance creation, not hoping "it will be set
        outside from master"
      
        NOTE ._v_blkdata = None means the block was not yet loaded and
        generally fits into logic how ZBlk operates and thus the change is ok.
      a354d36b
    • Kirill Smelkov's avatar
      bigfile/zodb: Do not waste DB space storing trailing zeros for a ZBlk · f696ce92
      Kirill Smelkov authored
      A lot of times data in blocks come shorter than block size and the rest
      of the memory page is zeros (because it was pre-filled zeros by OS when
      page was allocated).
      
      Do a simple heuristic and trim those trailing zeros and not store them
      into DB.
      
      With this change size of DB created by running bigfile and bigarray
      tests changes as following:
      
                  bigfile     bigarray
      
          old     145M         35M
          new      24K          6M
      
      Trimming trailing zeros is currently done with str.rstrip('\0') which
      creates a copy. When/if needed this could be optimized to work in-place.
      f696ce92
    • Kirill Smelkov's avatar
      bigfile/zodb: Make "set blkdata to be later saved to DB" explicit method of ZBlk · e0b00bda
      Kirill Smelkov authored
      - to keep things uniform with counterpart .loadblkdata()
      - so that master do not mess with ZBlk internals and works only through
        interface - this way it will be possible to use several
        kinds of ZBlk.
      e0b00bda
    • Kirill Smelkov's avatar
      bigfile/zodb: Clarify which stage .loadblkdata() and __{get,set}state__() do · a9072290
      Kirill Smelkov authored
      All those functions move data between DB and ._v_blkdata and only master
      then connects data to memory page. Make that fact explicit.
      a9072290