| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
| |
This makes sure that Gitlab::Metrics::RackMiddleware is added before
Gitlab::EtagCaching::Middleware.
|
| |
|
|
| |
This reverts commit cb10b725c8929b8b4460f89c9d96c773af39ba6b.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
An empty file in one of the instrumented directories will cause the app
to fail to start when metrics are enabled. Metrics aren't enabled by
default in development or test.
We could handle the empty file case explicitly, but a file could still
not define the constant it is expected to, so instead run the
initializer manually in a spec and check that it succeeds.
|
| | |
|
| |
|
| |
So we have a detailed view of what checks perform bad
|
| |
|
|
|
|
| |
Because this method is a Rails scope we have to instrument it manually
as regular the instrumentation methods only instrument methods defined
directly on a Class or Module.
|
| |
|
|
|
| |
This class does quite a few interesting things so let's instrument it so
we can see how much time is being spent in this class.
|
| |
|
|
|
| |
Since this isn't an ActiveRecord::Base descendant it wasn't
instrumented.
|
| |
|
|
|
| |
This allows us to see how much time is being spent in just parsing
HTML/XML documents.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit changes the way certain documents are rendered (currently
only Notes) and how documents are redacted. Previously both rendering
and redacting would run on a per document basis. The result of this was
that for every document we'd have to run countless queries just to
figure out if we could display a set of links or not.
This commit changes things around so that redacting Markdown documents
is no longer tied into the html-pipeline Gem. This in turn allows it to
redact multiple documents in a single pass, thus reducing the number of
queries needed.
In turn rendering issue/merge request notes has been adjusted to take
advantage of this new setup. Instead of rendering Markdown somewhere
deep down in a view the Markdown is rendered and redacted in the
controller (taking the current user and all that into account). This has
been done in such a way that the "markdown()" helper method can still be
used on its own.
This particular commit also paves the way for caching rendered HTML on
object level. Right now there's an accessor method Note#note_html which
is used for setting/getting the rendered HTML. Once we cache HTML on row
level we can simply change this field to be a column and call a "save"
whenever needed and we're pretty much done.
|
| |\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Resolve "Track the number of new Redis connections per transaction"
## What does this MR do?
Add a new metric counter, `new_redis_connections`, that contains the number of calls to `Redis::Client#connect` in the current transaction.
## Are there points in the code the reviewer needs to double check?
Not sure. I tested this in kind of a brute-force way:
1. Add a debugger in the monkey-patched `connect` method.
2. With metrics enabled, start the app and load a page.
3. The first Redis connection is created by `Rack::Attack` and isn't in a transaction, but still works fine.
4. The second Redis connection is within a transaction (the page load), and increments the counter.
5. If I reload the page, neither debugger is hit.
6. If I use a Redis client and do `CLIENT KILL` on my two existing clients, then reload the page, I get 3 and 4 again.
7. If I disable metrics collection, the debugger never gets hit.
## Why was this MR needed?
We may have a Redis connection leak somewhere, so adding metrics will let us track this.
## What are the relevant issue numbers?
Closes #18451.
## Screenshots (if relevant)
Hahaha nope, not relevant.
## Does this MR meet the acceptance criteria?
- [ ] [CHANGELOG](https://gitlab.com/gitlab-org/gitlab-ce/blob/master/CHANGELOG) entry added
- [ ] [Documentation created/updated](https://gitlab.com/gitlab-org/gitlab-ce/blob/master/doc/development/doc_styleguide.md)
- [ ] API support added
- [ ] Tests
- [ ] Added for this feature/bug
- [ ] All builds are passing
- [ ] Conform by the [style guides](https://gitlab.com/gitlab-org/gitlab-ce/blob/master/CONTRIBUTING.md#style-guides)
- [ ] Branch has no merge conflicts with `master` (if you do - rebase it please)
- [ ] [Squashed related commits together](https://git-scm.com/book/en/Git-Tools-Rewriting-History#Squashing-Commits)
cc @yorickpeterse
See merge request !4649
|
| | |
| |
| |
| |
| | |
Increment the counter `new_redis_connections` on each call to
`Redis::Client#connect`, if we're in a transaction.
|
| |/
|
|
|
|
|
| |
By default instrumentation will instrument public,
protected and private methods, because usually
heavy work is done on private method or at least
that’s what facts is showing
|
| |
|
|
|
| |
Now that this code is no longer part of Banzai::Filter it needs to be
instrumented explicitly.
|
| |
|
|
| |
This worker is called manually by `RepositoryCheck::BatchWorker` meaning it's not tracked automatically by the Sidekiq middleware.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
Make all of the nested constant instrumentation for core app code work
the same way, add mailer instrumentation, and add instrumentation to the
premailer gem.
|
| | |
|
| |\
| |
| |
| |
| |
| |
| | |
Instrument all service classes
This will help us see where (mostly) Sidekiq code is spending time.
See merge request !3675
|
| | |
| |
| |
| | |
Fixes gitlab-org/gitlab-ce#15162
|
| |/
|
|
|
|
| |
Instrumenting this class together with Gitlab::ReferenceExtractor causes
a StackError for some reason. Since Gitlab::ReferenceExtractor has most
of the interesting code we'll only instrument that class.
|
| | |
|
| |
|
|
|
| |
This allows us to track how much time of a transaction is spent in
dealing with cached data.
|
| | |
|
| | |
|
| |
|
|
|
| |
This adds instrumentation for the instance methods of
Gitlab::Git::Repository.
|
| | |
|
| | |
|
| |
|
|
|
| |
This isn't hugely useful and mostly wastes InfluxDB space. We can re-add
this whenever needed (but only once we really need it).
|
| |
|
|
|
|
|
|
|
|
|
|
| |
This particular setup had 3 problems:
1. Storing SQL queries as tags is very inefficient as InfluxDB ends up
indexing every query (and they can get pretty large). Storing these
as values instead means we can't always display the SQL as easily.
2. We already instrument ActiveRecord query methods, thus we already
have timing information about database queries.
3. SQL obfuscation is difficult to get right and I'd rather not expose
sensitive data by accident.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
This works by searching the raw source code for any references to
commonly used ActiveRecord methods. While not bulletproof it saves us
from having to list hundreds of methods by hand. It also ensures that
(most) newly added methods are instrumented automatically.
This _only_ instruments models defined in app/models, should a model
reside somewhere else (e.g. somewhere in lib/) it _won't_ be
instrumented.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The use of ActiveSupport would slow down instrumented method calls by
about 180x due to:
1. ActiveSupport itself not being the fastest thing on the planet
2. caller_locations() having quite some overhead
The use of caller_locations() has been removed because it's not _that_
useful since we already know the full namespace of receivers and the
names of the called methods.
The use of ActiveSupport has been replaced with some custom code that's
generated using eval() (which can be quite a bit faster than using
define_method).
This new setup results in instrumented methods only being about 35-40x
slower (compared to non instrumented methods).
|
|
|
This adds the ability to write application metrics (e.g. SQL timings) to
InfluxDB. These metrics can in turn be visualized using Grafana, or
really anything else that can read from InfluxDB. These metrics can be
used to track application performance over time, between different Ruby
versions, different GitLab versions, etc.
== Transaction Metrics
Currently the following is tracked on a per transaction basis (a
transaction is a Rails request or a single Sidekiq job):
* Timings per query along with the raw (obfuscated) SQL and information
about what file the query originated from.
* Timings per view along with the path of the view and information about
what file triggered the rendering process.
* The duration of a request itself along with the controller/worker
class and method name.
* The duration of any instrumented method calls (more below).
== Sampled Metrics
Certain metrics can't be directly associated with a transaction. For
example, a process' total memory usage is unrelated to any running
transactions. While a transaction can result in the memory usage going
up there's no accurate way to determine what transaction is to blame,
this becomes especially problematic in multi-threaded environments.
To solve this problem there's a separate thread that takes samples at a
fixed interval. This thread (using the class Gitlab::Metrics::Sampler)
currently tracks the following:
* The process' total memory usage.
* The number of file descriptors opened by the process.
* The amount of Ruby objects (using ObjectSpace.count_objects).
* GC statistics such as timings, heap slots, etc.
The default/current interval is 15 seconds, any smaller interval might
put too much pressure on InfluxDB (especially when running dozens of
processes).
== Method Instrumentation
While currently not yet used methods can be instrumented to track how
long they take to run. Unlike the likes of New Relic this doesn't
require modifying the source code (e.g. including modules), it all
happens from the outside. For example, to track `User.by_login` we'd add
the following code somewhere in an initializer:
Gitlab::Metrics::Instrumentation.
instrument_method(User, :by_login)
to instead instrument an instance method:
Gitlab::Metrics::Instrumentation.
instrument_instance_method(User, :save)
Instrumentation for either all public model methods or a few crucial
ones will be added in the near future, I simply haven't gotten to doing
so just yet.
== Configuration
By default metrics are disabled. This means users don't have to bother
setting anything up if they don't want to. Metrics can be enabled by
editing one's gitlab.yml configuration file (see
config/gitlab.yml.example for example settings).
== Writing Data To InfluxDB
Because InfluxDB is still a fairly young product I expect the worse.
Data loss, unexpected reboots, the database not responding, you name it.
Because of this data is _not_ written to InfluxDB directly, instead it's
queued and processed by Sidekiq. This ensures that users won't notice
anything when InfluxDB is giving trouble.
The metrics worker can be started in a standalone manner as following:
bundle exec sidekiq -q metrics
The corresponding class is called MetricsWorker.
|
