Milo: Use custom config caching layer

Milo: Use custom config caching layer

Use our own caching layer for luci config.  This allows for better failure conditions,
and also the configs can be viewed.

Some behavioral changes:
* GetSettings() should never error, just returns an unfilled config.

BUG=700055
Review-Url: https://codereview.chromium.org/2801463002
Committed: https://github.com/luci/luci-go/commit/a42d0e810765a5abf65fe3c4d8a5fe1b4645e12c

[Ryan Tseng, 2017-04-04 20:31:35]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-04 20:31:42]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-04 20:38:13]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-04 20:38:13]

Dry run: This issue passed the CQ dry run.

[Ryan Tseng, 2017-04-04 21:33:32]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-04 21:33:35]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[hinoka, 2017-04-04 21:36:41]

Description was changed from

==========
Milo: Use custom config caching layer

Use our own caching layer for luci config.  This allows for better failure
conditions,
and also the configs can be viewed.

BUG=700055
==========

to

==========
Milo: Use custom config caching layer

Use our own caching layer for luci config.  This allows for better failure
conditions,
and also the configs can be viewed.

Some behavioral changes:
* GetSettings() should never error, just returns an unfilled config.

BUG=700055
==========

[hinoka, 2017-04-04 21:36:41]

hinoka@chromium.org changed reviewers:
+ nodir@chromium.org

[hinoka, 2017-04-04 21:36:49]

hinoka@chromium.org changed reviewers:
+ hinoka@chromium.org

[chromium-reviews, 2017-04-04 21:37:07]

Demo: https://luci-milo-dev.appspot.com/admin/configs

On Tue, Apr 4, 2017 at 2:36 PM, <hinoka@chromium.org> wrote:

> https://codereview.chromium.org/2801463002/
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[commit-bot: I haz the power, 2017-04-04 21:40:36]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-04 21:40:37]

Dry run: This issue passed the CQ dry run.

[nodir, 2017-04-04 23:12:00]

On 2017/04/04 21:37:07, chromium-reviews wrote:
> Demo: https://luci-milo-dev.appspot.com/admin/configs
> 

this returns an internal server error for me

[nodir, 2017-04-04 23:12:08]

https://codereview.chromium.org/2801463002/diff/20001/milo/api/resp/status_st...
File milo/api/resp/status_string.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/api/resp/status_st...
milo/api/resp/status_string.go:1: // Code generated by "stringer
-type=Status,ComponentType,Verbosity"; DO NOT EDIT.
did stringer change?

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
File milo/appengine/common/config.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:34: const ID = "service_config"
this looks as common.ID for other packages, which is abstract. Rename to
ServiceConfigID?

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:44: // Data The binary proto of the config.
s/The/is the/ ?

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:72: // The service config does not exist, just
return an empty config
i think it does exist, but it is broken

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:79: settings.Swarming = nil
what if you add more fields there later? it is safer to do settings =
config.Settings{Buildbot:&config.Settings_Buildbot{}}

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:92: logging.Infof(c, "Loaded config entry from
%s", msg.LastUpdated.Format(time.RFC3339))
this functions is going to be used a lot, sometimes multiple times in a single
http request. This will cause multiple "Loaded config entry ..." logs in a
single request.

You could remove the logging from GetCurrentServiceConfig, or you could use
https://godoc.org/github.com/luci/luci-go/common/data/caching/proccache here to
cache the config in process memory for a short period of time

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:96: // UpdateServiceConfig fetches the service
config from luci-cfg
sometimes this to refers to luci-config as luci-config and sometimes as
luci-cfg. Let's consolidate to luci-cfg

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:97: // and then stores the binary proto in
datastore.
remove "binary" because it is not only binary?

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:108: logging.WithError(err).Errorf(c, "Could not
load existing config")
general Go suggestion: handle an error only once. If you log it and return it,
the calling code may log it again => many log lines for one error. How about
return fmt.Errorf("could not load existing config: %s", err) and the calling
code can log it?

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:116: cs := cfgclient.CurrentServiceConfigSet(c)
cfgclient is a wrapper around
https://godoc.org/github.com/luci/luci-go/common/config, but this code does not
use any of the features that the wrapper provides. Maybe use
https://godoc.org/github.com/luci/luci-go/common/config directly?

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:128: if msg.Revision != "" && msg.Revision ==
meta.Revision {
comparison and put must happen in the same transaction

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:128: if msg.Revision != "" && msg.Revision ==
meta.Revision {
!= "" seems unnecessary. The revision returned by luci-config cannot be empty

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:141: msg.LastUpdated = time.Now().UTC()
logging the fact that revisions do not match is more important than logging that
revisions match

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
File milo/appengine/frontend/config.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/config.go:50: logging.WithError(servErr).Errorf(c,
"service update handler encountered error")
yeah, double logging

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/config.go:53:
h.WriteHeader(http.StatusInternalServerError)
return

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
File milo/appengine/frontend/expectations/frontpage-Basic_frontpage.html
(right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/expectations/frontpage-Basic_frontpage.html:26: 
one day we won't have these

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
File milo/appengine/frontend/frontpage.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/frontpage.go:37: "errors":    err.Error(),
this looks like one error, a string. Why it is called errors? If you want to
support multiple errors, maybe make a slice

[nodir, 2017-04-04 23:27:15]

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
File milo/appengine/common/config.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:96: // UpdateServiceConfig fetches the service
config from luci-cfg
On 2017/04/04 23:12:08, nodir wrote:
> sometimes this to refers to luci-config as luci-config and sometimes as
> luci-cfg. Let's consolidate to luci-cfg

i meant let's converge on luci-config (not luci-cfg)

[Ryan Tseng, 2017-04-05 20:43:22]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-05 20:43:30]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[hinoka, 2017-04-05 20:45:07]

https://codereview.chromium.org/2801463002/diff/20001/milo/api/resp/status_st...
File milo/api/resp/status_string.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/api/resp/status_st...
milo/api/resp/status_string.go:1: // Code generated by "stringer
-type=Status,ComponentType,Verbosity"; DO NOT EDIT.
On 2017/04/04 23:12:08, nodir wrote:
> did stringer change?

I guess so, it did this in another CL also.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
File milo/appengine/common/config.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:34: const ID = "service_config"
On 2017/04/04 23:12:08, nodir wrote:
> this looks as common.ID for other packages, which is abstract. Rename to
> ServiceConfigID?

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:44: // Data The binary proto of the config.
On 2017/04/04 23:12:08, nodir wrote:
> s/The/is the/ ?

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:72: // The service config does not exist, just
return an empty config
On 2017/04/04 23:12:08, nodir wrote:
> i think it does exist, but it is broken

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:79: settings.Swarming = nil
On 2017/04/04 23:12:08, nodir wrote:
> what if you add more fields there later? it is safer to do settings =
> config.Settings{Buildbot:&config.Settings_Buildbot{}}

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:92: logging.Infof(c, "Loaded config entry from
%s", msg.LastUpdated.Format(time.RFC3339))
On 2017/04/04 23:12:08, nodir wrote:
> this functions is going to be used a lot, sometimes multiple times in a single
> http request. This will cause multiple "Loaded config entry ..." logs in a
> single request.
> 
> You could remove the logging from GetCurrentServiceConfig, or you could use
> https://godoc.org/github.com/luci/luci-go/common/data/caching/proccache here
to
> cache the config in process memory for a short period of time

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:96: // UpdateServiceConfig fetches the service
config from luci-cfg
On 2017/04/04 23:27:15, nodir wrote:
> On 2017/04/04 23:12:08, nodir wrote:
> > sometimes this to refers to luci-config as luci-config and sometimes as
> > luci-cfg. Let's consolidate to luci-cfg
> 
> i meant let's converge on luci-config (not luci-cfg)

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:97: // and then stores the binary proto in
datastore.
On 2017/04/04 23:12:08, nodir wrote:
> remove "binary" because it is not only binary?

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:108: logging.WithError(err).Errorf(c, "Could not
load existing config")
On 2017/04/04 23:12:08, nodir wrote:
> general Go suggestion: handle an error only once. If you log it and return it,
> the calling code may log it again => many log lines for one error. How about
> return fmt.Errorf("could not load existing config: %s", err) and the calling
> code can log it?

Ok i'll just wrap it.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:116: cs := cfgclient.CurrentServiceConfigSet(c)
On 2017/04/04 23:12:08, nodir wrote:
> cfgclient is a wrapper around
> https://godoc.org/github.com/luci/luci-go/common/config, but this code does
not
> use any of the features that the wrapper provides. Maybe use
> https://godoc.org/github.com/luci/luci-go/common/config directly?

Ah ok.  I didn't know that existed, I like that better.

edit: turns out the impl isn't currently installed in the context.  I'll punch a
hole to grab it from the backend.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:128: if msg.Revision != "" && msg.Revision ==
meta.Revision {
On 2017/04/04 23:12:08, nodir wrote:
> != "" seems unnecessary. The revision returned by luci-config cannot be empty

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/common/c...
milo/appengine/common/config.go:141: msg.LastUpdated = time.Now().UTC()
On 2017/04/04 23:12:08, nodir wrote:
> logging the fact that revisions do not match is more important than logging
that
> revisions match

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
File milo/appengine/frontend/config.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/config.go:50: logging.WithError(servErr).Errorf(c,
"service update handler encountered error")
On 2017/04/04 23:12:08, nodir wrote:
> yeah, double logging

Acknowledged.  Doesn't log on the other side now.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/config.go:53:
h.WriteHeader(http.StatusInternalServerError)
On 2017/04/04 23:12:08, nodir wrote:
> return

Done.

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
File milo/appengine/frontend/expectations/frontpage-Basic_frontpage.html
(right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/expectations/frontpage-Basic_frontpage.html:26: 
On 2017/04/04 23:12:08, nodir wrote:
> one day we won't have these

one day... Yeah these are ugly :(

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
File milo/appengine/frontend/frontpage.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/frontpage.go:37: "errors":    err.Error(),
On 2017/04/04 23:12:08, nodir wrote:
> this looks like one error, a string. Why it is called errors? If you want to
> support multiple errors, maybe make a slice

Done.

[commit-bot: I haz the power, 2017-04-05 20:47:59]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-05 20:48:00]

Dry run: Try jobs failed on following builders:
  Luci-go Mac Tester on luci.infra.try (JOB_FAILED,
https://luci-milo.appspot.com/swarming/task/3558ad553b364e10)

[Ryan Tseng, 2017-04-05 20:50:46]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-05 20:50:50]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-05 21:07:34]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-05 21:07:35]

Dry run: This issue passed the CQ dry run.

[Ryan Tseng, 2017-04-05 21:12:12]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-05 21:12:15]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-05 21:16:46]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-05 21:16:47]

Dry run: Try jobs failed on following builders:
  Luci-go Linux Trusty 32-on-64 Tester on luci.infra.try (JOB_FAILED,
https://luci-milo.appspot.com/swarming/task/3558c7a4ad4ae510)
  Luci-go Linux Trusty 64 Tester on luci.infra.try (JOB_FAILED,
https://luci-milo.appspot.com/swarming/task/3558c7a4f1dcc210)

[Ryan Tseng, 2017-04-05 21:21:34]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-05 21:21:44]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Ryan Tseng, 2017-04-05 21:34:00]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-05 21:34:04]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[nodir, 2017-04-05 21:47:50]

lgtm % comments

please make sure that UpdateServiceConfig does not call logging functions

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
File milo/appengine/frontend/config.go (right):

https://codereview.chromium.org/2801463002/diff/20001/milo/appengine/frontend...
milo/appengine/frontend/config.go:50: logging.WithError(servErr).Errorf(c,
"service update handler encountered error")
On 2017/04/05 20:45:06, hinoka wrote:
> On 2017/04/04 23:12:08, nodir wrote:
> > yeah, double logging
> 
> Acknowledged.  Doesn't log on the other side now.

still does

https://codereview.chromium.org/2801463002/diff/120001/luci_config/server/cfg...
File luci_config/server/cfgclient/backend/backend.go (right):

https://codereview.chromium.org/2801463002/diff/120001/luci_config/server/cfg...
luci_config/server/cfgclient/backend/backend.go:69: // GetConfigInterface
returns the raw configuration interface for the backend.
s/for/of/

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
File milo/appengine/common/config.go (right):

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:88: if item, ok := proccache.Get(c,
ServiceConfigID); ok {
if 2 goroutines call GetCurrentServiceConfig at the same time, both will go to
datastore. Use
https://godoc.org/github.com/luci/luci-go/common/data/caching/proccache#GetOr...

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:118: return err
you said you will wrap it, but this is not wrapping and you still log the error

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:124: c, "could not unmarshal proto from
luci-config:\n%s", cfg.Content)
did you mean to wrap and return the error? currently it just logs and goes on

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:126: newConfig := ServiceConfig{ID:
ServiceConfigID}
nit: all fields ut but ID are set with assignment statements. Be consistent.
Consider assigning all fields but Data like you assign ID

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:130: logging.Errorf(c, "could not marshal proto
into binary\n%s", newConfig.Text)
this double logs. wrap and return without logging?

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:165: proccache.Put(c, ServiceConfigID,
newConfig, time.Minute*10)
note that this will affect only one of your many appengine instances. Maybe
remove this and reduce timeout to like 1min?

[commit-bot: I haz the power, 2017-04-05 21:51:43]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-05 21:51:44]

Dry run: This issue passed the CQ dry run.

[Ryan Tseng, 2017-04-05 23:27:39]

The CQ bit was checked by hinoka@google.com to run a CQ dry run

[commit-bot: I haz the power, 2017-04-05 23:27:44]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-05 23:42:51]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-05 23:42:52]

Dry run: This issue passed the CQ dry run.

[hinoka, 2017-04-06 03:34:34]

https://codereview.chromium.org/2801463002/diff/120001/luci_config/server/cfg...
File luci_config/server/cfgclient/backend/backend.go (right):

https://codereview.chromium.org/2801463002/diff/120001/luci_config/server/cfg...
luci_config/server/cfgclient/backend/backend.go:69: // GetConfigInterface
returns the raw configuration interface for the backend.
On 2017/04/05 21:47:50, nodir wrote:
> s/for/of/

Done.

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
File milo/appengine/common/config.go (right):

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:88: if item, ok := proccache.Get(c,
ServiceConfigID); ok {
On 2017/04/05 21:47:50, nodir wrote:
> if 2 goroutines call GetCurrentServiceConfig at the same time, both will go to
> datastore. Use
>
https://godoc.org/github.com/luci/luci-go/common/data/caching/proccache#GetOr...

Done.

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:118: return err
On 2017/04/05 21:47:50, nodir wrote:
> you said you will wrap it, but this is not wrapping and you still log the
error

Done.

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:124: c, "could not unmarshal proto from
luci-config:\n%s", cfg.Content)
On 2017/04/05 21:47:50, nodir wrote:
> did you mean to wrap and return the error? currently it just logs and goes on

Done.

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:126: newConfig := ServiceConfig{ID:
ServiceConfigID}
On 2017/04/05 21:47:50, nodir wrote:
> nit: all fields ut but ID are set with assignment statements. Be consistent.
> Consider assigning all fields but Data like you assign ID

Done.

https://codereview.chromium.org/2801463002/diff/120001/milo/appengine/common/...
milo/appengine/common/config.go:165: proccache.Put(c, ServiceConfigID,
newConfig, time.Minute*10)
On 2017/04/05 21:47:50, nodir wrote:
> note that this will affect only one of your many appengine instances. Maybe
> remove this and reduce timeout to like 1min?

Done.

[hinoka, 2017-04-06 03:35:06]

The CQ bit was checked by hinoka@chromium.org

[hinoka, 2017-04-06 03:35:07]

The patchset sent to the CQ was uploaded after l-g-t-m from nodir@chromium.org
Link to the patchset: https://codereview.chromium.org/2801463002/#ps140001
(title: "Review: Remove double logging")

[commit-bot: I haz the power, 2017-04-06 03:35:16]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-06 03:54:12]

CQ is committing da patch.

Bot data: {"patchset_id": 140001, "attempt_start_ts": 1491449706954100,
"parent_rev": "023113801f132887a4ce958b81802bb76c2d7148", "commit_rev":
"a42d0e810765a5abf65fe3c4d8a5fe1b4645e12c"}

[commit-bot: I haz the power, 2017-04-06 03:54:18]

Description was changed from

==========
Milo: Use custom config caching layer

Use our own caching layer for luci config.  This allows for better failure
conditions,
and also the configs can be viewed.

Some behavioral changes:
* GetSettings() should never error, just returns an unfilled config.

BUG=700055
==========

to

==========
Milo: Use custom config caching layer

Use our own caching layer for luci config.  This allows for better failure
conditions,
and also the configs can be viewed.

Some behavioral changes:
* GetSettings() should never error, just returns an unfilled config.

BUG=700055

Review-Url: https://codereview.chromium.org/2801463002
Committed:
https://github.com/luci/luci-go/commit/a42d0e810765a5abf65fe3c4d8a5fe1b4645e12c
==========

[commit-bot: I haz the power, 2017-04-06 03:54:19]

Committed patchset #8 (id:140001) as
https://github.com/luci/luci-go/commit/a42d0e810765a5abf65fe3c4d8a5fe1b4645e12c