IndexedDB: Reusing leveldb logs when opening database

IndexedDB: Reusing leveldb logs when opening database

This change activates a new open mode in leveldb which reuses logs & manifest
to reduce time to open database.

This affects only Indexed DB databases, not other leveldb's opened elsewhere.

BUG=402980, 395799
Committed: https://crrev.com/de701abea081c633b69c05f797bb9587e825c6bd
Cr-Commit-Position: refs/heads/master@{#308493}

[cmumford, 2014-12-13 18:23:34]

cmumford@chromium.org changed reviewers:
+ jsbell@chromium.org

[jsbell, 2014-12-15 06:37:48]

lgtm

[cmumford, 2014-12-15 16:04:14]

The CQ bit was checked by cmumford@chromium.org

[commit-bot: I haz the power, 2014-12-15 16:04:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/803603004/1

[cmumford, 2014-12-15 17:05:55]

The CQ bit was unchecked by cmumford@chromium.org

[cmumford, 2014-12-15 17:29:20]

jsbell@: I missed one important thing; the implementation of
leveldb::Env::NewAppendableFile leveldb::Env has one which returns a not
implemented error causing leveldb to fallback to the old impl.

Can you give it one more look?

[jsbell, 2014-12-15 17:41:23]

Needs tools/metrics/histograms/histograms.xml update and the new enum value
needs to be inserted at the end.

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
File third_party/leveldatabase/env_chromium.cc (right):

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
third_party/leveldatabase/env_chromium.cc:915: } else {
No need for else block, since previous returns.

(I know, this was copy/paste from previous NewXXXFile impls)

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
File third_party/leveldatabase/env_chromium.h (right):

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
third_party/leveldatabase/env_chromium.h:22: enum MethodID {
Isn't this enum logged to UMA, and therefore new entries must be appended?

[cmumford, 2014-12-15 18:08:35]

cmumford@chromium.org changed reviewers:
+ jar@chromium.org

[cmumford, 2014-12-15 18:08:36]

+jar for histograms.xml review.

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
File third_party/leveldatabase/env_chromium.cc (right):

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
third_party/leveldatabase/env_chromium.cc:915: } else {
On 2014/12/15 17:41:23, jsbell wrote:
> No need for else block, since previous returns.
> 
> (I know, this was copy/paste from previous NewXXXFile impls)

Done.

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
File third_party/leveldatabase/env_chromium.h (right):

https://codereview.chromium.org/803603004/diff/20001/third_party/leveldatabas...
third_party/leveldatabase/env_chromium.h:22: enum MethodID {
On 2014/12/15 17:41:23, jsbell wrote:
> Isn't this enum logged to UMA, and therefore new entries must be appended?

Thx. I think I forgot this last time too. Added a comment to this enum.

https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
tools/metrics/histograms/histograms.xml:58882: +  <suffix
name="NewAppendableFile" label="ChromiumEnv::NewAppendableFile"/>
Note: I put NewAppendableFile in the IndexedDBLevelDBPFEMethods section, but
NewWritableFile is in IndexedDBLevelDBErrnoMethods. I believe that
NewWritableFile is n the wrong section. I don't believe we should change that.

[jsbell, 2014-12-15 18:37:46]

code still lgtm, histogram errno enums may require some follow-up cleanup.

https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
tools/metrics/histograms/histograms.xml:58882: +  <suffix
name="NewAppendableFile" label="ChromiumEnv::NewAppendableFile"/>
On 2014/12/15 18:08:36, cmumford wrote:
> Note: I put NewAppendableFile in the IndexedDBLevelDBPFEMethods section, but
> NewWritableFile is in IndexedDBLevelDBErrnoMethods. I believe that
> NewWritableFile is n the wrong section. I don't believe we should change that.

Now that base::File etc is used everywhere, does the Errno stuff get used at
all?

[cmumford, 2014-12-15 18:40:55]

All issues addressed, just waiting on jar's review of histogram.xml

https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
tools/metrics/histograms/histograms.xml:58882: +  <suffix
name="NewAppendableFile" label="ChromiumEnv::NewAppendableFile"/>
On 2014/12/15 18:37:46, jsbell wrote:
> On 2014/12/15 18:08:36, cmumford wrote:
> > Note: I put NewAppendableFile in the IndexedDBLevelDBPFEMethods section, but
> > NewWritableFile is in IndexedDBLevelDBErrnoMethods. I believe that
> > NewWritableFile is n the wrong section. I don't believe we should change
that.
> 
> Now that base::File etc is used everywhere, does the Errno stuff get used at
> all?

Nope.

[cmumford, 2014-12-15 21:14:45]

cmumford@chromium.org changed reviewers:
+ jwd@google.com

[cmumford, 2014-12-15 23:25:39]

On 2014/12/15 18:40:55, cmumford wrote:
> All issues addressed, just waiting on jar's review of histogram.xml
> 
>
https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
> File tools/metrics/histograms/histograms.xml (right):
> 
>
https://codereview.chromium.org/803603004/diff/40001/tools/metrics/histograms...
> tools/metrics/histograms/histograms.xml:58882: +  <suffix
> name="NewAppendableFile" label="ChromiumEnv::NewAppendableFile"/>
> On 2014/12/15 18:37:46, jsbell wrote:
> > On 2014/12/15 18:08:36, cmumford wrote:
> > > Note: I put NewAppendableFile in the IndexedDBLevelDBPFEMethods section,
but
> > > NewWritableFile is in IndexedDBLevelDBErrnoMethods. I believe that
> > > NewWritableFile is n the wrong section. I don't believe we should change
> that.
> > 
> > Now that base::File etc is used everywhere, does the Errno stuff get used at
> > all?
> 
> Nope.

Pinging for review of histograms.xml

[jar (doing other things), 2014-12-16 00:22:33]

jar@chromium.org changed reviewers:
+ asvitkine@chromium.org, isherman@chromium.org

[jar (doing other things), 2014-12-16 00:22:34]

histograms.xml (and associated histogram naming) LGTM

[cmumford, 2014-12-16 00:25:01]

The CQ bit was checked by cmumford@chromium.org

[commit-bot: I haz the power, 2014-12-16 00:26:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/803603004/40001

[commit-bot: I haz the power, 2014-12-16 01:29:15]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2014-12-16 01:30:02]

Patchset 3 (id:??) landed as
https://crrev.com/de701abea081c633b69c05f797bb9587e825c6bd
Cr-Commit-Position: refs/heads/master@{#308493}